Hey,

In a few weeks a song will start playing in every grocery store, gas station and waiting room you walk into. For me it's "Feliz Navidad." I've used it as an example in my seminars for years. It comes on, I groan, I turn it off. (Sorry, José.)

Me turning it off doesn't stop it. It plays somewhere else, all season long. And every time it plays, whoever owns the rights to that song collects a royalty.

"Feliz Navidad" came out in 1970. Fifty years later, in December 2020, it reached the Billboard Hot 100 top 10 for the first time.1 Same song, still paying.

Now think about how your December works. You pick up the extra shift, the overtime, the holiday hours somebody else didn't want. The money comes in, the gifts and the gas and the plane tickets go out, and in January you start over from zero. Your money shows up when you do, and only then.

Book my complimentary diagnostic call →

I've spent 18+ years in real estate, focused on short-term rentals since 2016, and built a list of more than 10,000 short-term rental operators through CashFlowDiary. The pattern below shows up in every cycle.

• • •

Piles and streams

The first lesson in my Cashflow Foundation course is one line. Build streams of income, not piles of cash.

A pile is a stack of money: a paycheck, a bonus, a tax refund, a savings account. Piles are good to have. But once you start spending a pile it only gets smaller, and the only way to grow it again is to go back to work.

A stream refills on its own schedule. A royalty is a stream. So is rent.

Most of us were taught one way to make money. Show up, get paid once, repeat, and save whatever is left. In August 2026, Americans saved 4.1% of their after-tax income.2 That's about four dollars out of every hundred. A pile grows slowly at four dollars a hundred.

"A big stack of money, that's wonderful, but unless you have a system that continues to provide you cash on a monthly basis, you have a problem."

- J. Massey, Cashflow Foundation course

The faucet in your kitchen

I use the kitchen sink to explain this in the course.

You don't keep a barrel of water in the yard and haul a bucket inside every time you want a drink. You turn the faucet and water comes, because a pipe connects your house to a supply.

A paycheck is the bucket. You carry it home, you pour it out, you go get another one.

A stream of income is the pipe. Somebody has to lay it, and it needs upkeep. As I tell students, passive does not necessarily mean uninvolved. But once the pipe is in, the water keeps coming on the days you don't clock in.

Where rentals fit

I can't write a song like that one. Most of us can't. You don't need a hit record to own a stream.

Rental income is the stream I know. A furnished rental earns when someone sleeps in it, and people need a place to sleep most during the same weeks most of us are stuck at work. AAA projected 81.8 million Americans would travel at least 50 miles from home over Thanksgiving 2025, and 122.4 million over the year-end holidays.3 Every one of them slept somewhere.

A rental takes setup, either money or a landlord's yes, and real work up front. It may not fit your life, and it doesn't pay from day one. It's one way to lay a pipe, and it's the one I've spent the last ten years on.

A 15-minute exercise before the holidays

1. List every dollar that came in last month. Paycheck, side gig, refund, a gift from your mom. All of it.

2. Mark each one P or S. P if it stops when you stop showing up. S if it would still arrive if you were sick for a month.

3. Count the S's. Zero is a normal answer. It tells you where to start.

4. Name one pile you could turn into a pipe. You don't have to act on it this week. Write it down.

Back to that song

This year when "Feliz Navidad" comes on, I'll still reach for the dial. But I'll remember somebody is getting paid for it, every time, more than fifty years later. And I'll ask myself the question I'm asking you: what do you own that pays you while you're at the holiday table?

If the honest answer is "nothing yet," that's the conversation I'd like to have with you. Book the complimentary diagnostic call with the link below, and we'll look at your numbers together.

- J.

P.S. If someone you love works every holiday so everyone else can be off, forward this to them. If someone forwarded it to you, subscribe here so the next one comes straight to you. And when you're ready to talk it through, book your complimentary diagnostic call.

This is education, not financial, legal or tax advice. Rental income varies, takes work and capital or a landlord's approval, and is never guaranteed.

Ready for the next step?

Hey,

Sixteen people asked about one of our properties and never booked.

I used to treat that like weather. Somebody asks, somebody leaves, you answer the next message. Then we went back through the inquiry history on that property, counted every one that didn't book, and sorted them by reason.

Sixteen people. Two reasons. And one of them cost nothing to fix.

Neither reason ever showed up in a review or a complaint. Below are both of them, and the two-hour audit you can run on your own inquiries this week.

Book my complimentary diagnostic call →

I've spent 18+ years in real estate, focused on short-term rentals since 2016, built a list of more than 10,000 short-term rental operators through CashFlowDiary, and recorded 705 published CashFlow Diary™ podcast episodes breaking down the deals that work and the ones that don't. The pattern below shows up in every cycle.

• • •

The rule we made up

Some of those sixteen guests wanted two nights. Our minimum was three.

Nobody objected to the price. Nobody complained about the property. Our settings said three nights, so the booking system turned those guests down for us while we slept, and nobody on our side ever saw it happen.

We changed the minimum. It took a few minutes.

What bothers me is how invisible it was. Nobody ever wrote to say the minimum was too long. The rejection happened inside a settings page, one guest at a time, for months. Without the recount, that rule would still be turning people away today.1

On Airbnb that rule sits under Trip length in your calendar's availability settings, and you can set a different minimum for each check-in day.2 So the fix doesn't have to be all or nothing. You can loosen the minimum on the days where guests keep asking and keep it everywhere else.

Before you rush to copy us, one caution. A shorter minimum can mean more cleans for the same nights, a stranded single night, or a stay your city's rules don't allow. Check those first. Our three-to-two change fit one property. Your log decides yours.

The offer nobody could see

The rest of the sixteen wanted something smaller than the full weekend event we were selling. A small private gathering.

We can host that. The property can do it. But nobody could see it. There were no photos of a small event in the space, no page describing it, no words for it anywhere on the site.

So a guest asking about a small gathering was reading a website that only talked about big ones, looking at photos of one kind of use. And they decided, reasonably, this isn't the place.

We lost those bookings for a marketing reason.

"The property could do it. The website never showed it."

- J. Massey, CashFlow Diary podcast, Season 4, Episode 2

Your inquiry log is your product roadmap

Guests keep asking for things slightly off from what you advertise. When that happens, there are two honest answers.

Either you can't deliver it, the answer is no, and that's fine. Or you can deliver it and you never showed it, and then the lost booking is on you.

For our small gatherings, four pieces were missing, and the order matters: photos of the offer happening, a defined experience, a price, and confidence you can deliver it the way you described it.

The log changed how we brief a photographer too. Our last professional shoot gave us handsome pictures of the wrong story, because nobody handed the photographer the questions guests actually ask. The next shot list comes straight from those questions, on the same budget.

And we're not waiting for perfect. Every photo we own was taken in summer, and it's already fall, with winter next. Usable photos this season beat a flawless shoot that lands after the season is gone.

We caught ourselves arguing about food tastings and menu options for those small gatherings. That conversation didn't matter yet. No photo, no defined experience, no price, no sale.

This week's two-hour audit

Set aside two hours. Then:

1. Pull every inquiry from the last twelve months that didn't book. Platform messages, emails, texts, calls and form fills. Log each guest once, even if they asked three ways.

2. Write one line for each. What did they ask for that you didn't give them? Use their words where you have them, and mark what you're guessing.

3. Sort the lines into piles. Pile A is rules and settings. Pile B is offers you could deliver but never showed. Anything else goes in other or unknown, and you leave it there.

4. Fix pile A first. It's usually a settings change. Check your costs, your cleaner's capacity and your local rules, change one thing, and write down the date.

5. Turn pile B into a production list. Photos, defined experience, price, delivery confidence, one offer at a time, starting with the one that showed up most.

Leave guest names and phone numbers out of it. You need the request, not the person.

What goes on each line matters more than the tool. Write the inquiry date, the dates and nights they wanted, the group size and what the stay was for. Then the unmet request, in the guest's own words if you have them. Then one honest label: did the guest say it, or are you guessing? "We only need two nights" is evidence. Silence after you sent a price is a guess, so it goes in unknown. Last, the change you'll try, the date you made it and what happened next, including any new cleaning cost or setup work.

Then wait until the inquiries that came in after the change have had time to book or not, and compare bookings from inquiries against a similar stretch last year. Change one thing at a time, so you know what moved.

Paid members get the Lost-Inquiry Log worksheet with this edition. A notebook works too: one line per guest, written the same way every time.

Keep reading: The full #699 article: How to Audit Lost Airbnb Inquiries Before You Buy More Traffic

Common questions about lost inquiries

How many inquiries do I need before this is worth doing? There's no minimum. We learned enough from sixteen to change a setting and start a production list. A small count finds patterns worth testing. It can't prove a result.

Should I just lower my minimum stay? Only if your log shows guests asking for shorter stays and the shorter stay works for your cleaning costs, your team, your calendar and your local rules.2

What if guests never told me why they left? Log them as unknown. Look for repeated questions your listing doesn't answer, and fix those, without pretending you know why each person left.

Ready to find your own leak?

Sixteen inquiries, two reasons, one of them free. Block the two hours this week, sort your own log, and fix the first pile before you spend another dollar on traffic.

• • •

P.S. Bring three rough numbers to a complimentary one-on-one diagnostic call: what you net, your manager's cut and your booking-site fees. You'll leave with your first move either way.

- J.

Educational content, not financial advice. Results not typical.

Most rental businesses don't lose money on bad deals. They lose it on follow-up that never happened. On a complimentary one-on-one diagnostic call, we look at who you should already be following up with and which part of that an AI coworker could carry. If it isn't a fit, I'll tell you.

Book a complimentary diagnostic call →

Somebody said yes on Friday. The invoice is out. This is where the yes cools off, unless somebody follows up.

Hey,

Ten days ago I told you my first hire wasn't a person.

On September 24, AI coworkers started sending one-to-one messages to people already in my CRM. They read each person's record first and wrote to them like a human would. A short ask. One question. A real reply if they wrote back.

That was day one. This is what the next two weeks looked like.

Six people got an offer on a call. Three have paid. Three are still deciding. And the payment I'm proudest of came from the kind of work every small business drops first.

I've spent 18+ years in real estate, focused on short-term rentals since 2016, built a list of more than 10,000 short-term rental operators through CashFlowDiary, and recorded 703 published CashFlow Diary™ podcast episodes breaking down the deals that work and the ones that don't. So here are the receipts. The wins, what it cost, and the part that went wrong.

Paid reader tool: The 66-Vertical Decision Guide

Paid members, annual or monthly, get the 66-Vertical Decision Guide with this edition: the 66 kinds of guests who book short and mid-term stays, and a worksheet to pick the three that fit your market. The article continues below for everyone.

• • •

The receipts

Here's everything from September 18 through October 7.3

• 1,992 people got a personal first message, mostly by text

• 321 people wrote back

• 141 asked us to stop, and we stopped

• 19 sales calls booked, counting reschedules and follow-up calls

• 8 diagnostic calls held

• 6 people got an offer on a call

• 3 paid $997 each, so $2,991 from new clients

• 3 offers still open, worth $4,991

• 2 existing clients bought more in the same two weeks: $10,321

• $13,312 in sales, all told

• $1,306 spent on the AI work and the texting fees

Two things up front, because they matter.

Two of the three new buyers found me through this newsletter, not through a text. The AI coworkers didn't create those two leads. What they did was everything after.

And most of the money came from two people who were already clients. Nobody had to find them. Somebody had to follow up.

The follow-up nobody has time for

Here's the payment that sold me.

An operator I'll call P. had a great call with me on a Friday. He said yes on the call. I sent the invoice. Then nothing.

You know this moment. Somebody said yes, and the yes is cooling off. On a small team, this is where it dies. Nobody decides to let it die. Monday just shows up with new calls, new guests, a cleaner who didn't show, and the warm yes from Friday slides down the list.

That didn't happen this time.

The next afternoon, an AI coworker built P. a personal plan page. His own numbers. What the first 90 days would look like for his property. When the plan pays for itself. It went to him with a clear deadline.

He paid the next morning at 9:46.

Nobody on a small team has time to build one person a custom page on a Saturday afternoon. Nobody skips a new appointment to do it either. That's the whole point. This is the work humans push aside to chase the next new lead, and it's the work that turns a yes into a payment.

P.'s warm yes, step by step: call, invoice, silence, a personal plan page, payment at 9:46 the next morning.

"The money was in finishing with the people who already said yes."

- J. Massey

Context to strategy to done, in about 30 minutes

I tested how fast this could go with another prospect.

• 10:59 AM: I recorded a short voice memo about his situation

• 11:07 AM: the analysis of his situation was done

• 11:17 AM: his personal page was built

• About 11:30 AM: it was sent

• 12:10 PM: he opened it

That one hasn't turned into a sale. I'm telling you anyway, because speed isn't the point. The point is that the follow-up happened the same morning, while the person still remembered why they cared.

On your own, that page shows up next week if you're lucky. By then they've moved on.

Same morning, while they still remember why they cared.

A human keeps the important replies

The first thing every operator asks me is, "Won't that feel like spam?"

Here's how it's set up:

• Anything about pricing, complaints, or someone I know personally comes straight to me

• STOP means stop, right away, on every channel

• Every first message reads the person's actual history before it goes out

• Replies get checked every five minutes, from 8 AM to 11 PM Eastern, seven days a week

The AI coworkers handle the volume. I handle the moments that need a person.

Where every reply goes: routine replies to the AI coworkers, pricing, complaints and people I know to me, and STOP ends contact on every channel.

What went wrong

Now the bad news.

141 people asked us to stop. That's about 7 out of every 100 people we messaged. Our own line to pause and fix things is 2 out of 100. We blew past it.

When I dug in, the messages were fine. The list was the problem. A lot of those texts went to people who bought something from me years ago and hadn't heard from me since. Some of the dates in my CRM were import dates, not purchase dates, so a few texts said "you bought back in 2024" to people who didn't.

So we changed it:

• No more texts to past buyers who don't read the newsletter

• People who haven't heard from me in a while get an email first, not a text

• Every date in a message has to be a real purchase date

• Newsletter readers get a message on Substack first

If you take one thing from this edition, take this: the tool will do exactly what you point it at. Point it at people who know you and want to hear from you.

What it would cost with people

This is my estimate, not something I measured.

If a person spent about 3 minutes reading each record and writing each first message, 1,992 messages is about 100 hours of work. That's two and a half weeks for one full-time setter, before a single reply.

Appointment setters average about $16 to $24 an hour in the U.S.12 So that first pass alone is roughly $1,600 to $2,400 in wages. We spent $1,306 for all of it: the first messages, the replies, the reminders, and the plan pages.

And a person doesn't check replies every five minutes until 11 PM on a Saturday. Two people can't either without burning out.

Not counted in my $1,306: my CRM subscription, the server, and card fees. I'm leaving them out because I'd pay those with or without the AI coworkers.

How to start this week

You don't need 1,992 people. Start with the ones you already have.

1. Day 1: Pick one list. Past guests, people who asked about a property, or people who said "maybe later." One list only.

2. Day 2: Write the job. Who gets contacted, the one question you ask first, and what has to come straight to you.

3. Day 3: Write the stop rule. STOP means stop. Decide it before you send anything.

4. Day 4: Send to 20 people. Not 200. Read every reply yourself.

5. Day 5: Count two numbers. How many wrote back, and how many asked you to stop. If more than 2 in 100 said stop, fix the list before you send more.

6. Day 6: Find your warm yeses. Anyone who said yes and then went quiet gets a personal follow-up with their own numbers.

7. Day 7: Decide. Keep going, fix the list, or stop.

Ready to find the follow-up you're missing?

On a complimentary one-on-one diagnostic call, we look at your lists, your warm yeses, and the follow-up that keeps slipping. You leave knowing which part an AI coworker could carry and which part should stay with you. If it isn't a fit, I'll tell you.

Book a complimentary diagnostic call →

P.S. Hit reply and tell me: who said yes to you in the last month and then went quiet? I read every reply.

You're reading Cashflow Diary Direct: real numbers, no hype, from an operator who's been in real estate for 18+ years.

This is educational content, not financial, legal, or tax advice. Results are not typical. Small sample: 8 calls, 6 offers and 3 new clients over about two weeks. Follow texting and consent rules where you operate before you message anyone.

• • •

Ready for the next step?

Book your complimentary one-on-one diagnostic call

Hey,

If you keep correcting your AI coworker's messages or reports, test a different model on that same job before changing the whole workflow.

I picked Astra for the website. I picked Smart for both videos. For the credit analysis, I picked Ultra.

Those choices came from a blind comparison. I saw outputs labeled A, B and C, with the model identities hidden and the letters reassigned for each task.

I wanted capability and a good user experience. Price wasn't my first priority.

Then I saw the credits. Across the eight jobs, Astra used about 16.4% fewer than Ultra. That surprised me. Smart used fewer than either of them.

I loved doing this test.

If you're choosing a model for your AI coworker, you can borrow the method. Start with work you recognize and decide what you would use before revealing who made it.

• • •

What I chose across eight jobs

The comparison ran on October 4, 2026, inside Viktor. Each of three settings received the same task brief for each job, producing 24 task runs.1

The workers read from a live workspace and live sources, so retrieved data and context weren't guaranteed identical. I judged final deliverables; first drafts weren't scored separately.

The workspace called the settings Ultra, Smart and gpt-6-astra. Ultra was Opus 5.5 with high reasoning; Smart was Opus 5.5 with medium reasoning. Both used the same Opus base model.

GPT-6 Astra used its default reasoning setting, listed as medium in the workspace catalog. I'll call it Astra. These are the settings as tested, rather than a promise about what every account offers.

My choices split this way:

• Astra: client SMS, research answer, blog draft and website.

• Smart: decision-request message, social video and explainer video.

• Ultra: credit analysis.

My first choices across eight tasks: Astra four, Smart three, Ultra one. One judge; one run per setting per task.

I recorded a favorite in each group. I didn't create a complete first-through-third ranking, and those picks don't tell us whether a runner-up was close.

The split gave me a practical starting point. I approved Astra as my default, retained Ultra for deep work, and chose Smart for video builds. Heavy, multi-source numerical analysis goes to Ultra.

Those are working choices for my setup. Your guest communication, reporting or website work may produce a different set of preferences.

Credits per job answer a different question

For the same eight-task bundle, Ultra consumed 9,655 credits, Astra consumed 8,074, and Smart consumed 4,023.2

Credits for the eight-task bundle: Ultra 9,655; Astra 8,074; Smart 4,023. Task-thread usage only, excluding parent orchestration and judging.

Astra used about 16.4% fewer credits than Ultra, while consuming roughly twice Smart's credits.

The website I preferred came from Astra and used 2,518 credits. Ultra's website used 1,352; Smart's used 656.

My preferred website was the most credit-intensive of the three. The cheaper bundle didn't mean a cheaper result on every task.

Provider token rates price units of model input or output. A finished job may involve multiple model interactions and other work.

The recorded credits tell us what these task runs consumed inside Viktor. They aren't measured provider spending, a token count or a cash invoice.

This trial didn't measure token totals or tool-call counts. It can't explain the credit difference by saying one model “thought less” or “used fewer tools.”

For your own test, save both the preferred output and its usage. If you convert credits to dollars, label the plan rate and explain what the conversion represents.

A plan-rate allocation can help with budgeting. Calling it an extra charge would require evidence that the charge occurred.

Look at the work behind the choices

Some written outputs used private business details, so I'm sharing the website and video examples instead.

The website task was a single-page diagnostic-call landing page. It required a self-contained HTML file plus desktop and mobile screenshots.

Selected trial screenshot (desktop crop); unedited test output. The live version includes booking and production fixes.

A screenshot lets you inspect the design I preferred. It cannot establish booking performance, and this test didn't measure conversion rates.

For a business website, you would still check the booking path, mobile layout and accuracy of every claim before using it.

The two video tasks were narrower than the word “video” might suggest.

One brief called for a 15-second vertical social piece. The other called for a 40-second horizontal explainer. Both used on-screen text and locally rendered motion graphics.

An excerpt from Smart's selected vertical social video. Silent, locally rendered motion graphics.

An excerpt from Smart's selected explainer. Silent, locally rendered motion graphics.

Smart was my first choice for both.

Audio was excluded. The trial tested no voices, music, actors or generated footage, so it can't tell you which setting would make the best narrated video.

The on-screen lending details are unedited test output, shown as design examples, not lending guidance.

I also can't offer a speed ranking. Runs shared computing resources, and video rendering involved a shared lock, waiting and rerendering.

In your own review, inspect the artifact your team would receive. Open the page on a phone. Watch the clip through its ending. Check the numbers in the report.

Run a small comparison you can repeat

You don't need eight tasks to begin. Choose a small set of recurring jobs with outputs you can inspect safely.

1. Choose the work and define “usable.” Pick a recurring task, then write its acceptance criteria before running it. For a guest message, that might mean accurate dates, appropriate tone and no unsupported promises. Remove private names and account details from test material.

2. Keep the brief and access consistent. Give each setting the same instructions, source material and tools. Save the model identifier and reasoning setting. A different brief or missing file can make the comparison difficult to interpret.

3. Hide identity and cost while judging. Have a helper shuffle the outputs into A, B and C. Keep the answer key separate. Write your preference and a short explanation before revealing the model. Allow “none is usable” as an answer.

4. Reveal usage and record the repair work. Compare task credits after choosing. Note factual errors, missing requirements and edits you would need. A favorite can still contain a serious mistake. Include failed attempts in the record.

5. Repeat before expanding the assignment. Try new examples of the same task, reshuffle the order and, where useful, ask another qualified person to judge. Recheck after a model or workflow changes.

You can start a worksheet with six fields: task, setting, pass or fail, blind preference, credits and required edits.

Keep safety separate from preference. If an answer invents a financial figure or exposes private information, mark that failure even if you like its presentation.

For recurring work, add frequency. A task you run every morning deserves different attention from a one-time page build.

Try the method on one recurring task

Write down what a usable result must get right, then judge the outputs before revealing settings or credits. Paid and founding readers also receive an editable scorecard and cost worksheet.

Related reading:

Common questions about this test

Did Astra win overall?

It received four of my eight first choices. That describes my preferences in this trial. With one run per setting per task and one human judge, the result doesn't establish a general quality ranking.

Should I pick the lowest-credit setting?

Use your acceptance criteria first. Smart had the lowest bundle usage, but I preferred other settings for five jobs. Your decision also depends on the consequences of an error and the work required to repair it.

Will the same setting win next time?

That remains untested. Repeat runs could produce different outputs and choices. Save the briefs, settings and results so you can compare another round instead of relying on memory.

Choose a task to improve in your business

Start with a recurring job you currently have to correct. Write down what the finished work must get right before changing its model.

Book your complimentary one-on-one diagnostic call

- J.

P.S. Keep your first test small enough that you can inspect every output yourself.

For educational purposes only. Results are not typical. Past performance does not guarantee future results.

Ready for the next step?

Two months ago I asked if you own your guest list. If the honest answer is still "not really," a complimentary one-on-one diagnostic call maps where your bookings depend on someone else's platform and what you can own outright. If it isn't a fit, I'll tell you.

Book a complimentary diagnostic call →

Two months on, the list on the porch is still the part you control. Editorial illustration, not a traffic forecast.

Hey,

On July 30 I sent you "Google Stopped Sending Traffic. Do You Own Your Guest List?" Two months later, I went back through every number in it. Two claims didn't hold up, one example had a math error, and a few new things happened that matter if you run short-term rentals.

The main point held. Most Google searches still end without a click to any website: 68% in the U.S. from January through April 2026, according to SparkToro's analysis of Similarweb data.1 Nothing since July has pushed that the other way.

So this edition covers three things: what I'm correcting, what's new, and a tighter version of the 30-day guest-list build.

I've spent 18+ years in real estate, focused on short-term rentals since 2016, built a list of more than 10,000 short-term rental operators through CashFlowDiary, and recorded 704 published CashFlow Diary™ podcast episodes breaking down the deals that work and the ones that don't. The pattern below shows up in every cycle.

Paid reader tool: The 30-Day Owned List Tracker

Paid members, annual or monthly, get a one-page tracker for the 30-day build in this edition, with the two numbers to check at the start and at day 30. The article continues below for everyone.

• • •

What I'm correcting from July

I'd rather correct these myself than have you find them.

The July 10 claim. I wrote that on July 10 Google made an AI answer the default response for every search. I couldn't find a Google source that says so. What Google did announce, on May 19, was a rebuilt search box and Gemini 3.5 Flash as the default model inside AI Mode.2 Regular results are still there. Fewer people scroll to them.

The Gmail threshold. I wrote that Gmail and Yahoo cut the allowed spam-complaint rate from 0.3% to 0.1%. They didn't. Google's guidance has long been to stay below 0.1% and never reach 0.3%.3 The real change came in July, when Gmail's Postmaster Tools started warning senders who go over 0.1%.4 Gmail now tells you when you've crossed the line.

I also made a math error in the July list example, fixed in the math callout below. A few other July figures were older or looser than they should have been. The current, sourced versions are in the next section.

What changed in search since July

Three updates, and all of them point the same way.

• The click loss got a second test. The randomized study I cited in July (about 1,000 U.S. Chrome users) now reports that AI Overviews cut outbound clicks to websites by 39.8% on the searches where they appeared, up from the 38% I quoted.5 A separate preregistered field experiment, published in August, found that removing AI Overviews and AI Mode raised clicks to publishers, and that an AI Mode-only page cut them.6 Ahrefs' own data puts the drop in clicks to the top-ranking page at about 58% when an AI Overview shows.7

• Google now reports AI visibility. As of August 31, every site in Google Search Console can see how often its pages show up inside AI Overviews and AI Mode.8 If you run a direct-booking site, that's a free number you didn't have in July.

• Ranking and being cited are still different jobs. Ahrefs found that only about 12% of the pages ChatGPT, Gemini and Copilot cited also ranked in Google's top ten for the same question.9 A page-one ranking doesn't guarantee an AI assistant mentions you.

Travelers still ask for the best place to stay near your market. More of them get the answer before they reach your site.

Where guests hear about you keeps changing. Where you keep them doesn't have to. Illustrative, not data.

What changed on the platform side

Search isn't the only place the ground moved.

Airbnb's fee. Airbnb has moved hosts to a single host-only fee. Most hosts pay 15.5%, taken from the payout.10 Hosts who use property management software and aren't on the single fee yet switch on October 13, 2026.11 If you haven't looked at your payout since the switch, look this week. Every night a past guest books directly is a night that fee doesn't touch.

Austin's delist rule. Since July 1, 2026, platforms must show a city license number on each Austin listing and remove a listing within 10 days of a city delist notice.12 The city paused bulk notices for about six months after its licensing system launched on May 18, and plans to send them in phases, starting with properties that drew nuisance complaints.13 Six months from May 18 is mid-November. If you operate in Austin, that date matters more than July did. If you don't, the lesson carries over: a city can take your listing down on its own schedule.

The 30-day owned guest-list build

Same plan as July, with one number added. AirDNA's 2026 Midyear Outlook calls for a steady year: demand and available listings both up about 2.7%, occupancy averaging 57.4%.14 When supply and demand grow at the same pace, nobody gets a free tailwind, so the operators who keep their past guests have the advantage.

1. Week 1: Capture. Put an email signup on your direct-booking site, above the fold. Add the link to your WiFi landing page and your check-in guide. Only add it to platform messages if your platform's off-platform communication rules allow it; check them first.

2. Week 2: First send. One email with a story from your property, one local recommendation and one direct-booking offer. Send monthly at minimum, the same week every month. Keep it to people who opted in, so your complaint rate stays well under Gmail's 0.1% line.

3. Week 3: Borrowed audiences. Pitch three podcasts, local blogs or travel newsletters your guests already read. Make one ask on every appearance: join the list. Other people talking about you also gives AI answer engines more places to find and cite you.

4. Week 4: Measure two numbers. List size first. Then open Search Console's generative AI report and write down how often your site appeared in AI answers this month. Check both again in 30 days.

If the monthly email is the step you know you'll skip, Monday's edition, "Your First Hire Isn't a Person," walks through how I handed follow-up work to an AI coworker.

In July I put it this way:

"If a stranger can hear about you and have no way to join your list, you don't have marketing. You have exposure."

- J. Massey

Rented: ranking, clicks and listings someone else can change. Conceptual illustration, not data.

Owned: past-guest emails and a send date you choose. Conceptual illustration, not data.

Questions operators are asking

I read the July edition and didn't act on it. Am I too late? No. The guests who stayed with you this summer are the easiest names to ask first. Start with Week 1 and send your first email in October.

Should I stop working on SEO? No. Keep your site clear and quotable, and use Search Console's AI report to see whether it shows up in AI answers. Treat search as one source of guests, and make sure it isn't your only one.

Does email still reach the inbox? Yes, for senders who follow the rules. Gmail's line is still 0.1% reported spam, and it now warns you when you cross it.34 Short, story-first emails sent to guests who asked to hear from you rarely get near that line.

I only have one property. Is this worth it? Yes. A single property with a small list of past guests who rebook directly depends less on search and platform ranking than a larger portfolio that relies on them alone. Owning the list matters more than the number of doors.

Keep reading:

Ready to own your guest pipeline?

If you want this build fitted to your property (where to capture emails, what to send and which borrowed audiences fit your market), book a complimentary one-on-one diagnostic call. If it isn't a fit, I'll say so.

Book a complimentary diagnostic call →

• • •

P.S. If you spot something in this edition that doesn't hold up, reply and tell me. I'll check it against the source and correct it, the same way I did with July's.

You're reading Cashflow Diary Direct: real numbers, no hype, from an operator who's been in real estate for 18+ years.

This is educational content, not financial, legal, or tax advice. Results are not typical. Past performance does not guarantee future results. Confirm current local rules and platform policies before changing how you market or book your rental.

Ready for the next step?

When a guest pays $1,000 for a stay, how much of it do you keep?

If you can't answer to the dollar, this is for you. I made a 3-minute video that shows where the money usually goes on one common setup: about $155 to the booking site, $200 to a 20% manager, and about $645 left before cleaning, repairs or empty nights.

▶ Watch the 3-minute video

It also covers the four numbers that decide what a rental keeps: how many people find you, how many book, how many come back, and how much of every stay you keep. Get about 10% better on each, and together that's about 46% more. That's an illustration, not a promise. Your property decides the real number.

Want to know which of the four is your biggest gap? Book a complimentary one-on-one diagnostic call. It takes 25 minutes. You bring three rough numbers, and you leave with one clear next step. If it isn't a fit, I'll tell you.

Pick a time →

- J.

Educational purposes only. Your numbers will differ.

If you're starting your first rental on nights and weekends, follow-up is where deals quietly die. On a complimentary one-on-one diagnostic call, we look at who you should be following up with and what an AI coworker could carry for you. If it isn't a fit, I'll tell you.

Book a complimentary diagnostic call →

The same table at 11pm and 7am. The follow-up doesn't care which one it is.

Hey,

Most people starting out picture the order of things like this.

First you get the property. Then the money comes in. Then, someday, you hire help.

I followed that order for years. On Thursday, September 24, I watched it flip.

We needed follow-up help. We didn't hire anyone. By that night, AI coworkers had sent 198 personal messages, handled the first replies, and put a call on my calendar. I took the call.

I've spent 18+ years in real estate, focused on short-term rentals since 2016, trained more than 10,000 operators through CashFlowDiary, and recorded 704 published CashFlow Diary™ podcast episodes breaking down the deals that work and the ones that don't. Here's exactly what happened, what it would have cost with people, and how you'd write the same job for your first property.

Paid reader tool: The First Hire Job Card

Paid members, annual or monthly, get a fillable worksheet that turns one follow-up job into written instructions for an AI coworker. The article continues below for everyone.

• • •

What happened in one day

We needed an appointment-setting team: someone to reach out one-to-one, answer replies, follow up, keep the CRM clean, book calls, and tell me who needs me.

We didn't hire anyone. We built it with AI coworkers in one day.

By that night:

• 198 people had gotten a personal text or email (173 texts, 25 emails). Not a blast. One person, one message.

• 17 had replied

• 3 asked us to stop (1.5%), and we stopped

• The first call was on my calendar

The part I keep thinking about: it was doing the work while we were still building it. Messages went out in the morning. By afternoon, we were adding the follow-up rules. By evening, the booked call had its own prep page, so I walked in knowing who I was talking to and what they'd already said.

The AI credits for the sales build and day one came to roughly $60 to $100. That's our estimate from the account data, not an invoice.

One day is one day. It's a baseline, not a forecast. I'll keep counting.

What the same week would cost with people

No knock on people here. I've hired setters, and good ones earn their pay. This is just the math a beginner never gets shown.

• An appointment setter earns about $16 to $24 an hour in the US before commission. PayScale puts the average at $16.08.1 ZipRecruiter data puts it at $24.26.2 Indeed's current figure is $20.07 an hour, plus about $4,400 a year in commission.3

• A freelancer to set up the CRM, pipelines and automations runs about $25 to $150 an hour on freelance marketplaces, per a 2026 pricing guide from Marketing Nativo. A fixed-scope build commonly lands around $300 to $2,500.4 GHL Prime puts agency setups at around $1,500 and up.5

• Then you recruit, train and wait. Often days or weeks before the first message goes out.

We did it for about the price of a nice dinner, and it started the same morning.

Why this matters more if you're just starting

Nobody says this to beginners: the hard part of your first rental is the follow-up, not the property.

The landlord who says "maybe next month." The investor who wants to see your numbers. The property manager who never called back. The deal is in the follow-up, and most beginners are doing it alone, at night, after a full-time job.

I've watched this for years. Someone gets a "maybe" on a Tuesday, means to follow up Friday, and by Friday the week has eaten them. Three weeks later the landlord signed with somebody else. Nobody said no. The follow-up just never happened.

That's the job AI coworkers are good at. What we built is a set of written instructions that runs the same way at 7am and at 11pm:

• reach out to landlords and investors one at a time

• follow up on day 4, day 10, and every 30 days

• keep a record of every conversation

• hand you the replies that need a human

Old order: property, then money, then help. New order: write the job, hand it to an AI coworker, take the calls yourself, get the first door. Conceptual illustration, not a results claim.

How to write the job tonight

You don't need to be technical for this part. You need a notepad and about an hour. This is the same shape as the job we wrote on September 24.

1. Who you contact. One list, one kind of person. Landlords with a unit that fits, or investors you already know. Not everyone at once.

2. The first message is a question. Something like "Are you open to a conversation about your unit on Oak Street?" No pitch, no attachment, no offer.

3. The follow-up schedule. Day 4, day 10, then every 30 days until they answer or say stop. Write the actual words for each one.

4. What comes to you. Pricing, complaints, anything legal, and anyone you know personally goes straight to you. The AI coworker doesn't answer those.

5. STOP means stop. One "stop," "no" or "remove me" ends it, on every channel, permanently.

6. Log everything. Every message goes on that person's record, so you can see the whole conversation before you pick up the phone.

7. One number per week. Pick one: calls booked. Check it the same day every week.

That's the whole job card. If you can explain it to a new hire, you can hand it to an AI coworker.

What we did NOT let the AI do

This matters, because nobody wants to be the person spamming their market.

• No pitch in the first message. The first message asks a question and offers nothing

• Anything about pricing, complaints, or people I know personally comes to me

• STOP means stop, permanently, across every channel

• Every message is logged on that person's record, so nothing is ever "I think we texted them"

Three people asked us to stop on day one. All three came off the list that day, on every channel. That's the rule working, and I'd rather lose a contact than keep someone who asked to go.

The guardrail split we used: routine follow-up goes to the AI coworker; anything that needs judgment comes to you. Conceptual illustration.

The real lesson

Most AI stories are about doing the same task over and over. This one is different: the AI coworkers helped build the department, and the department was working while it was being built.

If you're starting out, that changes the order:

• Write down the job (who you're contacting, what you ask, when you follow up)

• Hand it to an AI coworker

• Spend your own hours on the calls it books

You don't need to be able to afford a team before your first door. You need the team to get your first door.

Questions people are asking

Isn't this just spam with extra steps? It can be, if you do it badly. That's why the rules come first. One person, one message. The first message is a question. STOP is honored on every channel, right away. On our first day, 3 of 198 people asked us to stop, and all 3 came off the list. If you text anyone, follow texting consent rules and your platform's policies first.

Do I need tech skills? You need to be able to write clear instructions. The hard part is deciding who you contact, what you ask, and what comes back to you. That's thinking, not coding.

What does it cost, and which tools? We used AI coworkers connected to our CRM, which sends the texts and emails and keeps the records. Our AI credits for the build and first day were roughly $60 to $100, by our own estimate. Your cost depends on your tools and your volume. You'll pay for a CRM and texting either way, whether a person or an AI coworker does the follow-up.

Want to see what this looks like for your first property?

If you want help writing the job for your market (who to contact, what to ask, what should come straight to you), book a complimentary one-on-one diagnostic call. If it isn't a fit, I'll say so.

Book a complimentary diagnostic call →

• • •

P.S. Tonight, write down five people you've been meaning to follow up with and the one question you'd ask each of them. That list is your first job card. Everything else is just deciding who sends it.

You're reading Cashflow Diary Direct: real numbers, no hype, from an operator who's been in real estate for 18+ years.

This is educational content, not financial, legal, or tax advice. Results are not typical. Past performance does not guarantee future results. If you text or email people, follow texting consent rules (including the TCPA), anti-spam laws, and your platform's messaging policies, and honor every opt-out.

Ready for the next step?

Your slowest month deserves its own pricing plan. If you price the off-season the way you price peak season, you either leave nights empty or fill them at rates that don't cover the dip. On a complimentary one-on-one diagnostic call, we map where your calendar is leaking. If it isn't a fit, I'll tell you.

Book a complimentary diagnostic call →

An empty calendar can put an attractive rental under pressure. Editorial illustration, not a property forecast.

Hey,

If your market has an off-season, plan for it before the calendar slows. A quieter month is a reason to investigate, not an automatic instruction to cut every rate.

A pricing plan that fits peak months may not fit quieter weeks. When bookings slow, a blanket discount is tempting. But filling a night below its incremental cost adds work without improving cash flow. Start with the calendar and your costs, not a percentage picked under pressure.

I'd focus on three decisions: review prices week by week, test an additional guest segment before you need it, and measure the booking pace. These are planning steps, not a promise that every empty night can be sold.

I've spent 18+ years in real estate, focused on short-term rentals since 2016, trained more than 10,000 operators through CashFlowDiary, and recorded 704 published CashFlow Diary™ podcast episodes breaking down the deals that work and the ones that don't. Here's the planning process I'd use to examine a slower season.

Paid reader tool: Off-Season Test Card

Paid members, annual or monthly, get a six-page worksheet to document a pricing test, check booking contribution and set a stop rule. The article continues below for everyone.

• • •

The Off-Season Is an Unpriced Signal

Airbnb says its Smart Pricing tool adjusts nightly prices based on demand, using factors about the listing and its area. Hosts can set minimum and maximum prices and override individual dates.1 That is a tool you can use, not evidence of a universal seasonal occupancy swing. Build the plan around your property's actual booking history.

The failure mode is treating the whole off-season as one flat block instead of a curve. Within a four-month "slow" stretch there are still relatively stronger weeks — a holiday, a local event, a school break — sitting inside it, priced the same as the dead weeks around them.

Airbnb also lets hosts set custom prices for selected nights and offer weekly or monthly discounts. Its guidance says pricing and availability settings work together; offering a monthly discount requires allowing longer stays.2 Use those controls to test a specific change. They do not guarantee more revenue. Compare net contribution, not occupancy alone.

Fix the granularity first. Everything else in this issue builds on that.

A calendar that charges the same rate on the slowest Tuesday and the busiest Saturday of the month isn't a pricing strategy. It's a rate you picked once and stopped looking at.

- J. Massey

Reprice on a Real Calendar, Not a Gut Feeling

Start with up to two years of your own booking data, if available. Track occupancy, average daily rate (ADR), and lead time: the gap between booking and arrival. Compare booking pace at the same point before arrival. Lead time is a clue, not a test that isolates price from demand.

Run the four checks before you touch a rate — then test selected dates inside a cost limit you set first. Illustrative planning sequence, not a projection.

If a week is behind its usual booking pace, check comparable total prices, listing visibility, restrictions, and local events before changing rates. A small price test may help you learn; a different guest segment may be worth testing too. Neither low occupancy nor lead time alone tells you why guests have not booked.

Build the Second Demand Source Before You Need It

Consider demand beyond vacation travel: relocations, traveling professionals, or local-business visitors. Treat each as a hypothesis. Ask whether those guests need your location, layout, parking, and stay length. A segment's name is not proof that demand exists in your market, or that your property is a fit.

Start outreach before the season turns so you have time to learn. Before accepting longer stays, check local tenancy rules, your lease, insurance, and platform terms with the appropriate professionals. A longer stay does not bypass restrictions. Set a rate only after accounting for utilities, service costs, and the nights you may displace.

The Four-Week Build (Run This Before Your Next Off-Season)

1. Week 1: Review available occupancy, ADR, and lead-time data. Mark slower weeks and write down which price, visibility, or guest-segment question to test.

2. Week 2: Rebuild your calendar's seasonal pricing rules around actual weeks, not one flat off-season block.

3. Week 3: Stand up one non-tourist demand channel — extended-stay rate, local-business outreach, or an events-calendar tie-in.

4. Week 4: Set a repricing trigger (a lead-time threshold or occupancy floor) so the calendar adjusts itself instead of waiting for you to notice a slump.

Approach A — one number applied to every week of the slow season. Conceptual illustration, not data or a forecast.

Approach B — the same season read one week at a time, with a written floor before any test. Conceptual illustration, not data or a forecast, and not a recommended rate shape.

Common Questions From Operators

Isn't discounting the fastest way to fill an empty calendar? It may help if price is the obstacle, but it cannot guarantee bookings. Check visibility, restrictions, and comparable total prices first. Then test selected dates within your cost limits rather than discounting the whole season.

How far out should I build the second demand channel? Before the season turns, not during it. Extended-stay and local-business channels take weeks to warm up; starting them mid-slump means you're building the fix and living the problem at the same time.

What if I cannot find viable off-season demand? Avoid buying occupancy at a loss. Review whether quieter dates suit necessary maintenance or preparation, and budget for continuing fixed costs. An empty week still costs money; the question is which available choice limits the loss without creating a bigger problem.

Do I need new software to run this? Check the exports and reports you already have. If lead time is missing, calculate the days between booking and check-in from reservation records. A spreadsheet can hold the comparison. Buy another tool only after identifying a specific data or execution gap.

Ready to Stop Guessing at Your Off-Season?

If you want a second set of eyes on your calendar, book a complimentary one-on-one diagnostic call. We'll look at which weeks are price-sensitive and which need a different guest entirely. If it isn't a fit, I'll say so.

Book a complimentary diagnostic call →

• • •

P.S. — Pick three slow weeks from last year. Compare their booking pace, total prices, and restrictions with nearby alternatives where data is available. Write down one change to test and the cost limit you will not cross. Do not ask one number to explain the whole calendar.

This is educational content, not financial, legal, or tax advice. Results are not typical. Past performance does not guarantee future results. Confirm current rules and platform policies before changing your pricing or booking settings.

Ready for the next step?

Zero bookings on a listing priced under market. The rate wasn't the problem — the checkout screen was. On a strategy call we'll map where your own listings are quietly losing guests before they book. No pitch. You leave with the diagnostic.

Book your strategy call →

The Catskills tiny home at the center of this case study — 100 private acres, direct waterfront access, and zero October bookings before the fix.

Hey,

Picture a listing that's cheap, ranks well, and still won't book. That's what a Catskills-area tiny home looked like on September 15th: priced under the market's 25th percentile, sitting at the top of search, zero of 31 October nights on the calendar.

Every instinct says drop the price further. That instinct would have been wrong.

I've spent 18+ years in real estate, focused on short-term rentals since 2016, trained more than 10,000 operators through CashFlowDiary, and recorded 704 published CashFlow Diary™ podcast episodes breaking down the deals that work and the ones that don't. This is one of those breakdowns — and it's the clearest one I've seen this year for what actually causes a “good listing” to sit empty.

• • •

The Listing That Broke the Obvious Playbook

An automated employee I run — an “Occupancy Growth Manager” — checks every listing I operate five times a day and week: a 7am gap scan, a Monday scoreboard, a nightly drift check, a Thursday demand watch, a Wednesday listing review. On September 12th, it flagged the tiny home at roughly 8% occupancy through the whole summer.

The normal move here is: cut the rate. So for three straight days, that's what the scan tested. Three straight days, the answer came back the same: nothing left to cut. The rate was already under the 25th percentile of the market. Cutting further wasn't the lever.

That's what pushed the search somewhere else — the guest's total bill, not the nightly rate. And there it was: fees ran 24-60% of what a guest actually paid at checkout.1 Linen fee, resort fee, extra-guest fee, stacked on top of an already-thin rate — plus zero reviews since a recent relisting.

The rate looked cheap. The checkout screen was expensive. Nobody was lying to the guest; the listing just quietly cost more than it advertised, and guests noticed before they ever messaged.

A calendar that looks priced right on the search page and still gets zero bookings isn't a pricing problem. It's a bill problem — and most hosts never check past the nightly rate to find it.

— J. Massey

What Changed, in One Day

On September 15th, three moves went in together:

Kill the junk fees. Linen, resort, and extra-guest charges dropped to zero. Only the cleaning fee stayed.

Raise the base rate on purpose. Base went from $141 to $150, floor from $87 to $99 — protecting margin now that the fee income was gone.

Discount inside the pricing tool, not the platform. A 10% cut went on all 31 open October nights inside the pricing engine itself, not as an Airbnb-side promotional discount.2 That distinction mattered: an Airbnb-side promo on this exact listing, that same week, misfired on a sister property and raised its price instead of lowering it.

Same nightly rate the whole time. Only the fee structure changed — and the listing went from zero bookings to beating market in four days.

The Four Days That Followed — and What It Cost

Sep 15 (before): 0 of 31 Oct nights booked, under 25th percentile vs. market, performance index ~0. Sep 17: 26.7% occupancy, 34.3% vs. market, performance index 0.78. Sep 19: 36.7% occupancy, 35.2% vs. market, performance index 1.04 — beating market.

Three new reservations landed in the first 48 hours. A listing that had zero bookings for the entire month went to four nights booked before the week was out.

Here's the part most case studies skip: the real bill.

Querying the automation's own account directly (not a guess) put the credit cost of this specific investigation and fix at roughly $66 — $22 of that is an exact, isolated number for the thread where the fix was decided and executed; the rest is a documented range because two pieces of the analysis ran inside a longer conversation that also covered unrelated work that week.

Compare that to what a person would have cost to catch the same thing: an analyst doing the same daily gap-scanning and rebuilding the guest's all-in bill against comps runs $1,450-$4,500 at consultant rates for the 17-30 hours the work actually took. A full-service property manager, charging 20-25% of revenue, would likely have trusted the pricing tool's own suggestion to cut the rate further — the wrong move.

That's roughly a 35:1 cost ratio in the automation's favor. And the revenue at stake wasn't small: an estimated $2,300-$2,700 of October income was about to go unbooked on one small unit before the fix landed.

One honest caveat: the bookings that followed landed inside a week with real demand of its own — a college homecoming weekend, a reunion, fall foliage season. Treat the fix as a contributor alongside that demand, not the sole cause of it.

The One-Quarter Version, If You Run More Than One Door

1. Week 1: instrument the check. Whatever's currently “priced right” on paper: pull the guest's all-in total, not just the nightly rate, and compare it to two comps.

2. Week 2: separate the levers. If occupancy is weak and the rate is already competitive, stop touching the rate. Look at fees, photos, and reviews instead — the guest already saw the price and still said no.

3. Week 3: discount inside the tool, never the platform. A platform-side promo changes what the platform thinks your price is. A pricing-tool discount changes what the guest pays without moving your floor.

4. Week 4: put a number on what it would have cost you to miss it. Not to brag — to know whether the check paid for itself. If it didn't, you're checking the wrong thing.

Common Questions

Why not just lower the nightly rate again? Because it was already tested and ruled out. Three straight days of “nothing left to cut” told the automation the rate wasn't the lever — the total guest bill was.

Is a 35:1 cost ratio typical? No. It's specific to this listing, this fix, and this week's demand. Treat it as one documented example, not a universal multiple.

Could I catch this manually? Yes, if you check every listing's all-in guest total against comps on a schedule and act on what you find. Most operators don't have the hours to do that below a few hundred doors — that's the actual gap this closes.

Ready to See What Your Own Listings Are Quietly Costing You?

Start your free STR Systems Diagnostic — no pitch on the call, you leave with the diagnostic either way.

Book your strategy call →

• • •

P.S. — The tiny home in this story sits on 100 private acres with direct waterfront access — the kind of listing that looks like it should sell itself. It didn't, until someone (or something) checked the actual bill a guest pays, not just the number on the search results page. That's usually where the real leak is.

Disclosure: Results are not typical. Past performance does not guarantee future results. Educational purposes only — not investment, legal, or tax advice.

Today an AI employee did $683 worth of work for $59.76. If you run a short-term rental business, the same gap is probably sitting in stages of your guest journey nobody's staffing yet. No pitch — you leave with the diagnostic.

Start your diagnostic →

Today's receipt: $59.76 in credits versus $683 as an employee-day, $1,025 as a contractor-day.

Hey,

Today an AI employee on my team did roughly twenty hours of work.

Content edits, SEO fixes, email replies, deliverability monitoring, financial reconciliation — the kind of work that used to require a coordinator, a bookkeeper, and a marketing assistant, all pulling separate shifts.

The bill for that work: $59.76.

If I'd hired an employee to do it, same tasks, same day: $683. A contractor: $1,025.

It's a line-item receipt from today, itemized task by task.

Each line is cross-checked against 2025-2026 salary data for its role — SEO editor, email specialist, bookkeeper, operations coordinator — thirteen roles in total.1

I've spent 18+ years in real estate, focused on short-term rentals since 2016. I've trained more than 10,000 operators through CashFlowDiary, and recorded 704 published CashFlow Diary™ podcast episodes breaking down the deals that work and the ones that don't.

It's a receipt from my own business, today, not a theory I read about.

I'm telling you because most short-term rental operators are sitting on the exact same gap. Most of them don't know it exists.

🔒 Quick question: do you know what your own busywork actually costs you — employee, contractor, or otherwise — versus what it would cost to run through an AI employee? Most operators don't. Paid and founding members get the worksheet below to find out in five minutes. If you'd rather keep guessing, that's your call — just don't be surprised by the number later.

Upgrade to unlock →

• • •

The seven stages most operators never staff

Every guest who books your property moves through seven stages: discovery, evaluation, booking, pre-arrival, stay, post-stay, and advocacy — the point where they either become a repeat guest or a referral source.

I wrote about this leak in detail in The 7 Stations: Why Your Guests Leak Before They Book. This piece picks up where that one left off.

Ask yourself: how many of those seven stages does your business work, versus hope it goes fine?

Most operators are fully staffed — meaning a human or a tool is actively doing something — for maybe two of the seven: booking (the OTA handles it) and the stay itself (cleaning, check-in).

The other five stages run on autopilot, or don't run at all.

Discovery is the algorithm's job, not yours. Evaluation is whatever your listing photos happen to say.

Pre-arrival is a canned message template, if that. Post-stay is a review request, sent once, forgotten.

Advocacy — the stage where a happy guest turns into a referral or a direct rebooking — usually doesn't exist as a deliberate process at all. It happens sometimes, by accident.

Operators call that “word of mouth” instead of what it is: an unstaffed job that occasionally does itself anyway.

You don't have a marketing problem. You have five unstaffed stages between the two you're paying attention to.

— a line I use on every strategy call, because it's true every single time.

Most operators fully staff two of the seven guest-journey stages. The other five run on autopilot.

Your tools cover less than you think

Hospitable, Guesty, whatever PMS you're on — they're built to manage the stay.

Calendars, messaging templates, cleaning schedules. Good tools, doing one job well.

None of them write the content that gets a guest to discover you outside of Airbnb's algorithm, or nurture a past guest into a direct rebooking six months later.

None of them turn a five-star review into a referral asset, either. That work sits in the gap between “property management” and “guest relationship.”

Almost nobody staffs it, because hiring a person for it doesn't pencil out at the volume one or two properties produce.2

Those tools do exactly what they were built to do — manage the stay.

The problem is operators mistake “I have a PMS” for “I have my guest journey covered.”

Ask a PMS to write a post-stay email sequence that turns a five-star reviewer into three referrals, and it can't. That was never its job. It was built to manage the stay, not to grow the business around the stay.

Occupancy is not the only place you make money

Ask an operator how they made money last month, and almost every answer is a variant of “occupancy rate.”

That's the only stage most people have built a business model around.

But advocacy, pre-arrival upsells, post-stay retention offers, direct-booking content that routes around the 15-20% OTA cut — those are money on the table.

They stay on the table because doing them by hand doesn't scale past a handful of units.

A second revenue line requires someone to write it, send it, track who responded, and follow up with the ones who didn't — every week, for every property.

At two units, that math never works with a human doing it manually. It's a staffing math failure, plain and simple.

It's been true for as long as short-term rental has existed as a category. It hasn't been true for the last two years.

The receipt

This is where today's numbers matter.

The work an AI employee did today — real tasks, not hypothetical ones — would have cost $683 as an employee-day or $1,025 as a contractor-day. It cost $59.76.

That's an order of magnitude, not “a little cheaper.”

And it means the stages of the guest journey that never penciled out — the content, the nurture sequences, the review-to-referral loop — start penciling out the moment the labor cost drops that far.

What this looks like

An AI employee here doesn't mean a chatbot that answers FAQs.

It means installing something that runs a defined slice of the guest journey, and teaching your team how to run and supervise it. In practice, that install looks like:

1. Audit which of the seven stages your business currently staffs (most operators find it's two).

2. Pick one unstaffed stage with the clearest dollar impact — usually advocacy or pre-arrival.

3. Install the AI employee on that one stage first — content, follow-up sequences, or review-to-referral routing.

4. Train your team to supervise and correct it, not operate it by hand.

5. Measure the receipt — track what that stage costs to run in credits vs. what a human hire would cost, the same way I did today.

6. Expand to the next stage once the first one is running clean.

So that's the work we do.

Not selling you software and walking away — installing the employee, training you and your team on it, and handing over a system that keeps producing after we leave the room.

Ready to see what an AI employee could take off your plate?

Book a strategy call and we'll map which guest-journey stage is costing you the most right now.

Start your diagnostic →

• • •

P.S. — If you want the actual spreadsheet — real cron-by-cron itemization, real 2026 salary data per role, employee cost vs. contractor cost vs. credit cost — reply to this email and I'll send it over.

Paid subscribers get the fill-in-your-own-numbers version above.

This edition is for operators building a real business, not a hobby listing. If someone forwarded you this, subscribe to get the next one.

Who this is for

Real estate investors and short-term rental operators who want the real numbers — not theory. I’ve spent years running Cashflow Diary and coaching STR operators across 17 countries, and this newsletter is where I share what’s actually working (and what isn’t) in real time. If you own or want to own income property — especially short-term rentals — you’re in the right place.

What free subscribers get

Twice a week, I send a real edition: real deals, real numbers, real operating decisions from my own portfolio and from operators I coach. No filler, no “content marketing” fluff. Every edition also includes at least one practical, usable idea you can apply immediately, whether or not you ever pay for anything.

What paid subscribers get

Paid unlocks the parts that take the real work off your desk: the tools, templates, and step-by-step implementation blocks built into every edition — the stuff you’d otherwise have to build yourself or pay a consultant for. Think of it less like a subscription and more like Prime for your STR business: the value keeps showing up, keeps changing, and keeps compounding as long as you’re in. Monthly, annual, and founding members all get full access — there’s no second-class paid tier.

What’s coming

More tools tied directly to the numbers that move your business — Leads, Conversion, Frequency, Margin — plus deeper looks inside real deals as they happen, not after the fact.

Get started

• New here? Subscribe free to get every edition.

• Want the tools and implementation blocks too? Upgrade to paid any time.

Glad you’re here.

— J.

Hey,

This is the final part of a 7-part series on a real revenue rebuild for a luxury short-term rental. [Part 1](https://cashflowdiary.substack.com/p/your-property-has-1-revenue-line) mapped the revenue lines. [Part 2](https://cashflowdiary.substack.com/p/mpi-the-one-number-that-tells-you) gave you MPI. [Part 3](https://cashflowdiary.substack.com/p/the-7-stations-why-your-guests-leak) mapped the stations. [Part 4](https://cashflowdiary.substack.com/p/the-940night-setting-nobody-checks) drained the leaks. [Part 5](https://cashflowdiary.substack.com/p/125-points-the-direct-booking-math) built the channels. [Part 6](https://cashflowdiary.substack.com/p/the-food-question-how-culinary-takes) broke the ceiling. This one answers the only question left: what does running it cost? Details anonymized. Numbers real.

Six issues of this series have been education. This one is a decision frame — and one disclosure up front: my company sells one of the options below. I'll price every alternative first, tell you which ones are cheaper than us, and when you should take them.

Because here's what almost nobody selling to property owners will say: some of the alternatives are the right answer. Which one depends on a single number — your gross revenue — and there's a crossover point where the cheap option becomes the most expensive thing you own.

I've spent 15+ years in this space, trained more than 10,000 operators through CashFlowDiary, and recorded 237+ podcast episodes breaking down the deals that work and the ones that don't. This is the issue where I price my own industry — mine included.

• • •

What running the engine actually takes

Strip away every sales pitch and the revenue engine from this series — seven stations, multiple channels, content, capture, upsells, events — is 17 distinct service lines totaling about 135 hours a month. That's 0.84 of a full-time person. Lean, it's 100 hours; at a full editorial standard, 200.

That number is the honest starting point for every option below. Anyone quoting you a price without telling you the hours is hiding the ball.

Every alternative, priced

Option 1 — The percentage manager. Full-service STR management runs 20–35% of gross for a property like this — the luxury band clusters at 25–35%, and one major manager's own guide admits the industry ranges "10% to 50%." No upfront cost, and on a struggling property that feels safe. Hold this one; the crossover section below is about it.1

Option 2 — Buy the pieces yourself. Seven services à la carte from US freelancers and agencies, at published rates: website build $3,000–8,000 · CRM implementation $4,000–8,000 · SEO retainer averaging $3,209/mo across 439 surveyed providers · content at $250–399 per article · social management $1,000–4,000/mo. Bought at the lowest defensible rates, the bundle runs about $29,000 in year one. At mid-market, $76,000. And the part no line item shows: nobody coordinates them. Seven vendors, seven briefs, and you just became the marketing director.

Option 3 — Staff the actual scope at US rates. Price the real 135 hours with US freelancers, line by line, and it totals about $15,700 a month — roughly $188K a year. This is what "just hire good people" costs when the hours are counted instead of guessed.

Option 4 — The boutique agency. Strategy retainers commonly run $3,500–7,500/mo; the omnichannel versions that also do outbound run $12,000–25,000/mo. You're buying editorial judgment, which is real — but for a single property, most of what that retainer delivers is volume you can get for a tenth of the price.

Option 5 — A full-service flat engagement (what my company sells): roughly $8K/month all-in, about $96K a year after setup. Sits between Option 2's chaos and Option 3's payroll.

The crossover: when cheap becomes the most expensive option

Here's the math that decides between Option 1 and everything else.

A percentage manager at 25% costs you a quarter of gross, forever. A flat engagement costs ~$96K a year no matter what. Set them equal: the lines cross at about $384K of gross revenue.

Below that, the percentage manager is cheaper — and if your property will never exceed it, take that deal. I mean it. A $150K property pays a 25% manager $37,500 and a flat engagement makes no sense.

But this series has been about a property engineered past that line. At the engagement's Year-3 all-lines projection of $878K gross, the 25% manager costs $219K — every year, growing with every dollar the plan earns. The flat engagement still costs $96K. The gap is $123K a year, and it never comes back.

The percentage model taxes success. The flat model taxes commitment.

Which one you want depends entirely on which side of $384K you're building toward — and after six issues, you know exactly which side this property is aimed at.

The break-even, in nights

Abstract percentages hide what matters, so here it is in the property's own unit: nights.

At the case study's measured $694.50 net per night, the $9,500 setup pays for itself in about 14 incremental nights — once. The monthly cost is covered by 11–12 incremental nights a month. Against that: a measured occupancy gap of half the market (Part 2), a blocked best-seller unblocked (Part 4), two new channels (Part 5), and $710 a booking in upsells (Part 6). The plan doesn't need to work perfectly. It needs to find twelve nights.

So that's the market, priced without flattery: a percentage that's right for small and wrong for ambitious, a bundle nobody coordinates, a payroll number, an agency premium, and a flat engagement in the middle.

What's below the line is the part I'd want if I were deciding: the full 17-line catalog with hours, which 45% of it automates, the three-role staffing model if you'd rather build it than buy it, what this looks like as a business at 1, 5, and 10 clients — and exactly what our version includes.

The 17-line catalog, with hours

The full stack, grouped by station, with monthly hours at operating tempo:

Read more

Hey,

This is Part 6 of a 7-part series on a real revenue rebuild for a luxury short-term rental. [Part 1](https://cashflowdiary.substack.com/p/your-property-has-1-revenue-line) mapped the revenue lines. [Part 2](https://cashflowdiary.substack.com/p/mpi-the-one-number-that-tells-you) gave you MPI. [Part 3](https://cashflowdiary.substack.com/p/the-7-stations-why-your-guests-leak) mapped the stations. [Part 4](https://cashflowdiary.substack.com/p/the-940night-setting-nobody-checks) drained the leaks. [Part 5](https://cashflowdiary.substack.com/p/125-points-the-direct-booking-math) built the channels. This is the expansion issue. Details anonymized. Numbers real.

Run every play in this series on the $2M Hudson Valley estate — fix the settings, seal the stations, build the channels — and the lodging math tops out around $155K–$188K a year.

The property was bought against a $300K expectation. Lodging cannot get there. Not with better marketing, not with perfect pricing. The nights × rate arithmetic simply runs out of nights.

Here's what closes the gap, modeled year by year on this property: a culinary program contributing $37,450 net in Year 1, $80,467 in Year 2, $123,485 in Year 3. Stretch lodging at $187,941 plus culinary Year 3 equals $311,426 — the first version of this plan where $300K is arithmetic instead of aspiration.1

I've spent 15+ years in this space, trained more than 10,000 operators through CashFlowDiary, and recorded 237+ podcast episodes breaking down the deals that work and the ones that don't. Revenue-line expansion is the difference between a rental and a hospitality business.

• • •

Why lodging has a ceiling and food doesn't

A 7-bedroom house has a hard constraint: 365 nights. Push occupancy and rate to their realistic limits and you've found the ceiling. Every lodging-only property has one; most owners just haven't computed theirs.

Food breaks the constraint because it doesn't need a night to sell. A supper club sells seats to people who never sleep over. A brunch sells to guests who already paid for the night. Catering sells to the wedding party staying elsewhere. Same asset, new meters running.

And it moves every lever at once. Part 1 readers remember L×C×F×M — leads, conversion, frequency, margin. A seasonal menu is a reason to return (F). A flagship dinner raises the take per stay (M). A ticketed event introduces the property to future lodging guests (L). Nothing else in the stack touches all four.

The four products, with honest margins

Modeled on this property against confirmed regional price anchors:

• Weekend brunch (guest add-on) — $65/head × 16 = $1,040 gross · $249 net · 24% margin

• Supper club (ticketed, incl. non-guests) — $125/head × 20 = $2,500 gross · $950 net · 38% margin

• Flagship custom dining — $166/head × 16 = $2,656 gross · $1,209 net · 46% margin

• Drop-off catering — $35/head × 30 = $1,050 gross · $464 net · 44% margin

The anchors are real, not hoped: private-chef bookings in this region realize an average of $166 per person. Farm dinners run $75–150 a seat. One famous regional resort charges $400 a seat for its chef's table — that's the ceiling, not the target.

Now the finding that surprised even the program's own designer: brunch — the launch product — has the *worst* margin of the four. A chef and a server cost nearly the same whether the ticket is $65 or $166, and brunch has the lowest ticket. Launching with brunch is still right — lowest complexity, easiest yes, fastest reviews. But sell it internally as the learning product, not the margin play.

The $710 that shows up before anyone cooks harder

Part 3 readers have seen this number — here's where it comes from. Put just two of these products in the guidebook's upsell store, at modest attach rates:

Brunch at 30% attach × $1,040 = $312 expected per booking. Flagship at 15% × $2,656 = $398. Roughly $710 of expected culinary revenue per lodging booking — against the $40–60 that generic add-ons like firewood and early check-in produce.

That's the compounding trick: the culinary program makes the lodging bookings worth more, before a single outside ticket is sold.

What the $2.5M neighbor actually does with food

June Farms — the regional benchmark this series keeps returning to — looks like a food business. Restaurant, events, farm dinners, weddings.

Here's what their own wedding packages reveal: food is not included in any of them. Every event routes to an outside caterer. June Farms sells the site, the alcohol, and the staff — and takes 20% service on the whole bill while carrying zero kitchen risk.

The most successful hospitality property in this region doesn't run its own kitchen for events. It monetizes food without owning the food operation.

Read that again if you're an owner who's been told "add food" means "become a restaurant." That model — site + alcohol + margin on someone else's cooking — is available to any estate, and it's the template we'll price out below the line.

The multiplier hiding in the driveway

One more piece of the $311K math. The case study property has a second unit — a one-bedroom tiny home — currently producing almost nothing.

Plugged into the engine this series has been building — same channels, same capture, same upsell store, same guest list — it models at roughly $23,000 a year, for about $25 a month of incremental software cost. The engine is built once. Units plug into it. That's the quiet argument for doing this whole build properly: the second property, and the third, ride for nearly free.

So the expansion picture is real: four products with honest margins, a per-booking lift that starts immediately, a proven no-kitchen model next door, and a per-unit multiplier. What stands between an owner and all of it isn't cooking skill.

It's three compliance gates — and the consultant-grade advice about them is wrong often enough that this property's own program brief had to be corrected on four counts. That's below the line.

Gate 1 — The kitchen you're not allowed to use

Read more

Hey,

This is Part 5 of a 7-part series on a real revenue rebuild for a luxury short-term rental. [Part 1](https://cashflowdiary.substack.com/p/your-property-has-1-revenue-line) mapped the revenue lines. [Part 2](https://cashflowdiary.substack.com/p/mpi-the-one-number-that-tells-you) gave you MPI. [Part 3](https://cashflowdiary.substack.com/p/the-7-stations-why-your-guests-leak) mapped the stations. [Part 4](https://cashflowdiary.substack.com/p/the-940night-setting-nobody-checks) drained the free leaks. Now we build. Details anonymized. Numbers real.

The $2M Hudson Valley estate in this series sold 48 nights in its first 51 days of data — and paid its platform $6,052.59 in host service fees for them.

That's an effective 15.50%, verified to the penny across every booking. Not a listing mistake, not a setting. The structural price of renting 100% of your demand from one company.1

This issue is the math of paying less for the same guest — and the honest version, including where direct booking is oversold.

I've spent 15+ years in this space, trained more than 10,000 operators through CashFlowDiary, and recorded 237+ podcast episodes breaking down the deals that work and the ones that don't. Channel economics is where operators either build equity or rent it forever.

• • •

The fee you can't negotiate — and the one you can choose

If you run property-management software, Airbnb's single fee applies: 15.5%, all of it out of your payout, no opt-out. That's the baseline.

Now the alternative. A direct booking site through the property's own PMS — Hospitable Direct on its recommended plan — costs the host 3%, with the guest paying 4% and payment processing included.

Same guest. Same nights. Same house. 12.5 percentage points less to the middle layer — about $287 back on this property's average booking.

What 12.5 points is worth in a year

Two numbers, and the difference between them matters:

The ceiling: ~$9,800 a year. That's 12.5 points on this property's roughly $78K trailing-year gross — if every dollar shifted direct. It won't. Nobody's does. Anyone selling you that number is selling.

The realistic Year-1 blend: ~$3,600. A defensible first-year mix — half Airbnb, a fifth Vrbo, a fifth direct + Google, a tenth premium OTA — takes the effective commission from 15.5% to roughly 10.9%. That's the number to build a plan on, and it grows every year the direct share climbs.

The three flavors of "direct" — and the one that wins

Hospitable Direct comes in three models, and owners routinely pick the wrong one:

• Basic — host pays 1% + your own Stripe ≈ 3.9% · recovers 11.6 pts

• Premium split (the default) — host pays 3%, guest pays 4% · recovers 12.5 pts

• Premium host-only — host pays 7% · recovers 8.5 pts

The split model beats host-only by $32.10 per booking on this property — and it's the one that includes chargeback protection, guest vetting, and $5M damage coverage. On a 16-person, $4,000+ booking, those protections matter far more than they do on a studio. Cheaper and safer is rare. Take it.

The 0% channel hiding behind the 3% channel

Here's the part that turns one build into two channels: Google's vacation rental surface lists properties for free — no commission on referrals or bookings — but only if you have a working direct booking site to send travelers to.

The entry gates are minor: five-plus photos (use twenty), no watermarks, accurate calendar and pricing. The case study property had this channel already configured and inert — because its direct site was the broken "0 bedrooms" page from Part 4. Fix the page, flip the switch, and the only 0%-commission demand surface in existence starts working.

Why single-channel is a risk, not just a cost

The fee math gets the attention. The risk math should.

At 100% Airbnb, one algorithm change, one suspended listing, one policy shift is a 100% revenue event. Diversification is the fix, and the additions aren't equal:

Vrbo is the fee-efficient add — around 5% PMS-connected, and its platform data claims materially larger group bookings and longer lead times than Airbnb like-for-like, with a strong share of rural-market bookings. Take the exact platform-published percentages with salt; the direction is what matters. A large-group rural estate is Vrbo's home turf, and its ~48-day booking window directly attacks this property's biggest measured weakness: almost nothing on the books beyond 30 days.

Booking.com is a demand add, not a fee play. Around 15% commission plus payment fees — roughly Airbnb money, sometimes worse. You list there for reach and midweek international demand, not savings. Going in expecting a discount is how owners sour on diversification.

The wedding marketplaces are the sleeper. The Knot, WeddingWire, Zola. Not as a venue — as where the wedding party sleeps. Every regional venue that hosts a 200-person wedding creates 200 people who need beds nearby, and almost no estate publishes a page built for that search. More on the setup below the line.

So: 15.5% is the price of one channel. The blend costs 10.9% and falls every year. The question left is the one that actually stops owners — how do you set this up without becoming a web developer and a tax accountant?

Below the line: the build order, the two-layer site architecture, the tax registration nobody warns you about, and the legal documents that replace Airbnb's terms of service.

Read more

Hey,

This is Part 4 of a 7-part series on a real revenue rebuild for a luxury short-term rental. [Part 1](https://cashflowdiary.substack.com/p/your-property-has-1-revenue-line) mapped the revenue-line gap. [Part 2](https://cashflowdiary.substack.com/p/mpi-the-one-number-that-tells-you) gave you MPI. [Part 3](https://cashflowdiary.substack.com/p/the-7-stations-why-your-guests-leak) mapped the seven stations. This issue is the one you can act on today. Details anonymized. Numbers real.

Before I built a single thing or charged a dollar on this engagement, I ran a settings audit on the $2M Hudson Valley estate you've been following since Part 1.

The audit surfaced a conservative modeled $17,351 a year in recoverable revenue. Not from marketing. Not from new channels. From settings that were already wrong — and every fix was free.1

Part 3 ended with a rule: fix the leaks before you pour. This issue is the leak list. Three settings, each one checkable on your own account this afternoon.

I've spent 15+ years in this space, trained more than 10,000 operators through CashFlowDiary, and recorded 237+ podcast episodes breaking down the deals that work and the ones that don't. A settings audit is the first hour of every engagement I run — because it's where the free money hides.

• • •

Fix #1 — The minimum-stay rule blocking the best-selling product

The property required 3-night stays on Fridays and Saturdays.

Now read its own booking history. 5 of its 17 bookings were 2-night Friday–Saturday weekends — at $940 per night net, 35% above the property's average rate, adding up to 28% of all net revenue.

The highest-paying product the property had ever sold was the exact product its settings now refused to sell. A 2-night weekend request now hits a wall the owner didn't know existed.

The fix is one toggle: drop the Fri/Sat minimum to 2 nights — and raise the Fri/Sat rate, because the history says these guests pay a premium. That's roughly $9,400 of proven, already-demonstrated demand, unblocked in under a minute.

Fix #2 — A pricing tool configured for a different house

The property runs a dynamic pricing engine. The engine is fine. The numbers it was anchored to belong to a different property.

Base price: $600. The median listed rate for 7-bedroom properties in its own neighborhood: $959. Every price the algorithm generates is an adjustment applied to that base — so every night starts from a 37% discount to the segment and inherits the error.

Minimum price: $475. Roughly one forward night in five was pinned to that floor — and still not selling. A floor that low doesn't buy occupancy. It just lowers the average whenever a booking happens to land on it.

Maximum price: $1,900. On one peak fall weekend, a 9-bedroom comp listed at $2,063, a 7-bedroom comp at $2,593–2,893, and a brandless 5-bedroom log house at $1,840. The estate's ceiling sat below what smaller neighbors actually charge on peak dates. The best weekends of the year were capped by a number typed in months earlier.

The engagement's earlier research modeled this reconfiguration at roughly $22,000 a year. I'll be straight with you: that's a model, not a measurement — the mechanism is what's certain. A fifth of nights priced at a floor that isn't converting, and a base anchored 37% under the segment, is rate given away for nothing.

Fix #3 — The broken page teaching Google the wrong thing

The property's direct booking page was live and indexed — displaying "0 bedrooms."

That page occupied the best URL on the property's own domain. It was the first thing a search engine — and any guest doing their homework — learned about a 7-bedroom estate: that it apparently had no bedrooms and looked abandoned. Ten minutes to fix. It had been undermining every search result since launch.

And one more storefront problem in the same family: months after the owner bought the property, the old for-sale listing was still live on the major real-estate sites at just under $2M — the purchase price. Guests researching a luxury stay found a for-sale sign instead of a destination. One email to the listing agent closes it.

Why I'm giving you the fixes for free

Two reasons. First, the numbers overlap — the $9,400 and the ~$22,000 both push toward the same modeled scenario, which is why the conservative first-pass total is $17,351 and not the sum. Precision matters more to me than a bigger headline.

Second: these three fixes are the findings. What they have in common is the method that found them — reading the settings against the property's own booking history and its own segment, instead of against defaults. Your property's three fixes are almost certainly different from these three.

The methodology — how base, floor, ceiling, and the override rules actually interact, the five customizations that silently pierce your minimum price, the trap inside your pricing tool's own recommendation, and the 10-setting audit checklist — that's below the line.

How the four price controls actually interact

Your pricing tool has four controls that most owners treat as one. They behave differently, and two of them lie.

Read more

Hey,

This is Part 3 of a 7-part series on a real revenue rebuild for a luxury short-term rental. [Part 1](https://cashflowdiary.substack.com/p/your-property-has-1-revenue-line) mapped the revenue-line gap. [Part 2](https://cashflowdiary.substack.com/p/mpi-the-one-number-that-tells-you) gave you MPI, the number that measures it. This issue shows you where the leaks actually are. Details anonymized. Numbers real.

Last issue ended with a promise: the property's 15.5% platform fee was the cheapest of its problems.

Here's what I meant. When owners hear "revenue problem," they picture one leak — the commission. So they obsess over the fee while six other holes drain the boat.

The way to see all seven at once is to stop looking at your property and start looking at your guest. Every guest who ever pays you passes through seven stations. At each one, they either move forward or they're gone.

I've spent 15+ years in this space, trained more than 10,000 operators through CashFlowDiary, and recorded 237+ podcast episodes breaking down the deals that work and the ones that don't. The framework below is the one I use to diagnose every property I touch.

• • •

Discovered → Considered → Answered → Booked Direct → Anticipating → In Residence → Returning.

The 7-bedroom estate from Parts 1 and 2 — a luxury property on 100 acres in the Hudson Valley — has exactly one of these stations built. One. Someone finds the Airbnb listing. Everything else either leaks or never existed.

Walk the stations with me. Count how many you have.

Station I — Discovered

A guest who has never heard of you learns you exist.

For the case study property, this station is one channel: Airbnb. 100% of bookings, 100% of revenue, one algorithm's opinion between the owner and an empty calendar. That's why its MPI reads 0.50 — when a guest chose this area, half the time the property wasn't even in the room.

When this station is thin, nothing downstream matters. You can't convert a guest who never saw you.

Station II — Considered

The guest found you. Now they're comparing you against every tab they have open.

This is photos, listing copy, the first three lines, reviews — and whatever Google says when they search your property's name. The case study's direct page was live and indexed, displaying "0 bedrooms." Every guest who checked looked at what appeared to be an abandoned listing.

June Farms — the regional benchmark from Part 1 — plays this station at a different level: a Prime Video series and 102K Instagram followers doing the considering for the guest before they ever compare prices.

Station III — Answered

The guest has a question. Someone — or something — answers it, fast.

Here's the scenario that convinced me this station matters more than owners think. A corporate retreat buyer is up at 9pm researching venues for sixteen people. She sends a question through the site. She has no reservation — so the property's platform-native AI tools can't even see her. They only watch confirmed bookings. Her question sits unanswered overnight, and by morning she's booked somewhere that replied at 9:04.

That's not a hypothetical demand profile. In the case study's inquiry log, 22 of 54 inquiries were for groups of 7 or more — including 9 inquiries at 14–16 guests. The biggest checks in the pipeline are exactly the buyers who ask questions before booking.

An unanswered question isn't a missed message. It's a closed sale, reversed.

Station IV — Booked Direct

The guest books — and where they book decides who keeps the money.

The case study pays an effective 15.5% on every Airbnb booking, verified to the penny across 17 reservations. A direct booking on the property's own site costs about 3%. That's a 12.5-point recovery on every dollar shifted — on this property's average booking, $287 back per reservation, for the same guest, the same nights, the same house.1

Most owners never build this station because they think it's a website project. It isn't. It's a channel decision. More on the economics below the line.

Station V — Anticipating

The gap between booking and arrival — usually weeks of silence. That silence is the most expensive quiet in your business.

A guest who has already committed $4,000 to a weekend is the easiest person on earth to sell a $1,040 brunch or a private chef dinner. On the case study property, a digital guidebook with an upsell store projects to roughly $710 of added revenue per booking at modest attach rates — not from pushing harder, but from offering once, at the moment anticipation peaks.2

Generic add-ons (early check-in, firewood) run $40–60 per booking. The difference is having something real to sell. That's why Part 1 was about revenue lines.

Station VI — In Residence

The guest is in the house. This station isn't about service — it's about who you know is in the house.

A platform booking gives you one contact: the person who paid. On a 16-guest estate, fifteen people sleep under your roof whose names you will never learn — every one of them a person who has experienced the property and might book it, recommend it, or hold their company's retreat there.

Unless they join the WiFi. A WiFi capture portal turns one reservation into up to sixteen contacts instead of one. On a large-group property, that single mechanism is the difference between a list that grows by 20 names a year and one that grows by 300.

Station VII — Returning

The stay ends. Three assets should exist that didn't before: a review that ranks you, a contact list you own, and a guest who books again — without either of you paying a platform for the reintroduction.

The case study's measured repeat-guest rate: 7.1%. Not because guests didn't love it — six reviews, 5.0 average — but because nothing exists to bring anyone back. No list, no reason to return, no way to book that doesn't restart at Station I, full commission due.

The scorecard

Seven stations. The case study property, before this engagement: one built, six missing or leaking.

So here's the question that matters for your property: which stations do you have, which are leaking — and what, specifically, do you build at each one? There's a right tool for every station, a right order to build them in, and one strategy for Station IV that sounds backwards until you see the math.

That's below the line.

The tech stack, station by station

Read more

Hey,

This is Part 2 of a 7-part series on a real revenue rebuild for a luxury short-term rental. [Part 1](https://cashflowdiary.substack.com/p/your-property-has-1-revenue-line) mapped the revenue-line gap between the client's estate and its neighbors. This issue gives you the number that measures your own gap. Details anonymized. Numbers real.

Last issue I showed you a 7-bedroom estate with better amenities than almost everything around it — booking half of what its market books.

"Half of what its market books" isn't a feeling. It's a number you can pull for your own property in about ten minutes. Most owners have never looked at it.

It's called MPI, and it settles the argument every underperforming host is quietly having with themselves: is it my pricing, or is it something else?

I've spent 15+ years in this space, trained more than 10,000 operators through CashFlowDiary, and recorded 237+ podcast episodes breaking down the deals that work and the ones that don't. The pattern below shows up in every cycle.

• • •

MPI in plain English

Market Penetration Index = your occupancy divided by your market's occupancy.1

That's the whole formula. If comparable properties near you are 40% booked and you're 40% booked, your MPI is 1.0 — you're filling at the market's rate. The industry calls that "fair share."

Below 1.0, the market is booking and you aren't. Above 1.0, you're filling faster than the market — which usually means you're priced too low and buying occupancy you didn't need to buy.

One number. It doesn't care about your photos, your reviews, or your renovation budget. It only asks: when a guest chose this area, how often did they choose you?

The case study property's live numbers

Here's the estate from Part 1, straight from its pricing engine's market pulse:

• September — market 34% · property 17% · MPI 0.50

• October — market 23% · property 10% · MPI 0.43

• November — market 11% · property 10% · MPI 0.91

Read October again. For every 10 nights the neighborhood sells, this property sells 4.

November looks better only because almost nobody's November is booked yet — 11% market occupancy means the window hasn't opened. And it gets worse when you benchmark against real demand: this same market actually reached 59% last October. Measured against what the neighborhood realized a year ago, October's ratio is 0.17.

The pricing engine's own flag on this listing: "Your listing needs attention."

The diagnosis hiding inside the number

Here's where MPI earns its keep. Pair it with rate, and it tells you what kind of problem you have.

• Next 30 days — ADR $909 · RevPAR $212

• Next 60 days — ADR $882 · RevPAR $147

• Next 90 days — ADR $895 · RevPAR $109

• Next 180+ days — ADR $914 · RevPAR $66

The nightly rate holds near $900 in every window. RevPAR — revenue per available night — collapses from $212 to $66.

That one line changes what you spend money on. This is not a pricing problem. The property already has a pricing tool, correctly configured, doing its job. What it doesn't have is anyone generating demand.

A pricing tool cannot fix a demand problem. It can only discount into one.

MPI has two siblings

MPI measures demand. ARI (Average Rate Index) — your rate divided by your market's — measures price position. RGI (Revenue Generation Index) — your RevPAR divided by market RevPAR — is the bottom line, and RGI = MPI × ARI.2

You read them in order, because RGI alone tells you nothing about cause. An RGI of 1.0 could mean genuine parity — or you bought occupancy with discounts (high MPI, low ARI), or held rate and lost nights (low MPI, high ARI). Same score, opposite problems, opposite fixes.

The case study property reads: MPI 0.50 · ARI 0.93 · RGI 0.47. Rate is nearly at market. Occupancy is at half. The revenue shortfall is almost entirely demand.

One warning before you check yours: on some platform-connected listings, owner-blocked dates count as bookings. The case study's second unit showed 100% occupancy — with zero actual bookings. If your MPI looks suspiciously healthy, check whether your own blocks are propping it up.

So: your occupancy, divided by your market's occupancy. If the answer starts with a zero-point-anything below 0.9, the question stops being whether you have a gap and becomes what to do about it — which band you're in, which levers move first, and which of the traps in your pricing tool is quietly making it worse.

That's below the line. Including the three free fixes that surfaced $17K+ on this one property in a single afternoon.

How to fix your MPI

Pull the number, step by step

Read more

Hey,

This is Part 1 of a 7-part series. Over the next seven issues, I'm opening the file on a real engagement: a full marketing and revenue rebuild for a luxury short-term rental that was earning a fraction of what it should. Real data, real numbers, real mistakes — including some of mine. Details are anonymized to protect the client. The numbers are not.

I spent two days building competitive intelligence dossiers on every comparable property within 30 miles of a client's estate in rural upstate New York.

I read their rate cards. I pulled their booking calendars. I read the raw code on their websites to see what marketing tools they run. I counted every separate way each property takes money.

What I found changed how I think about this entire business.

I've spent 15+ years in this space, trained more than 10,000 operators through CashFlowDiary, and recorded 237+ podcast episodes breaking down the deals that work and the ones that don't. The pattern below shows up in every cycle.

• • •

The property that started this

The client owns a 7-bedroom luxury estate on 100 acres in the Hudson Valley. Sleeps 16 under one roof. Resort pool, home theater, underground wine cellar, gym, game room. A separate waterfront tiny home on the same land.

It was purchased for $2 million. It's earning $67K a year. One listing, on one platform, selling one thing: nights.

The pricing engine we use scores every property against its market. This one sits at an MPI of 0.50 — Market Penetration Index. Translation: it books half of what comparable properties book. Beyond 30 days out, the calendar is nearly empty while the market keeps filling.1

Here's what makes that number strange. On amenities, this property beats almost everything near it. The house is not the problem.

So I went looking at the neighbors.

What the neighbors are actually selling

The top-performing property in the region is a 120-acre farm about 20 minutes away called June Farms. You may have seen it — they have a Prime Video series and 102K Instagram followers.

Here's what most people don't see. June Farms is not a lodging business. Our modeled estimate, built from their published rates, confirmed volume anchors, and public pricing, puts them at roughly $2.57M a year across nine distinct revenue lines.2

Nightly lodging — the thing most hosts think of as the whole business — is 27% of it.

• Lodging (11 units) — $684K (27%)

• Weddings — $630K (25%)

• Restaurant + bar — $450K (17%)

• Day admission + ticketed events — $306K (12%)

• Private parties — $300K (12%)

• Merch + gift cards — $200K (8%)

Nine lines, grouped into six buckets there. Weddings alone out-earn every bed on the property. They charge $10 a head just to walk the farm — and that admission line converts strangers into future wedding couples and overnight guests.

Two more facts that should stop you cold. June Farms has never run a paid ad. Not one, ever — we verified it in the ad archives, which retain seven years of history. And their largest single unit sleeps six people.

They built a $2.5M business on structure, not spend.

The grid

Now widen the lens. I ran the same revenue-line count on every serious competitor in the region.

June Farms (120 acres): 9 lines. Weddings, restaurant, admission, ticketed events, parties, merch, retreats — plus the beds.

A 70-acre multi-building estate: 5 lines. Weddings, banquet hall, multi-unit lodging, full buyouts.

A 17-cabin architect-designed venue: 4 lines. Wedding buyouts, retreats, cabin weekends, event add-ons.

A 12,000 sq ft Georgian on 60 acres: 3 lines. Corporate offsites, retreats, celebrations.

The client's estate: 1 line. Nothing beyond nights.

Read that bottom row again. The property with arguably the best single-house amenity set in the entire comparison — the pool, the theater, the wine cellar, the 100 private acres — sells exactly one thing. And it books half of what the market books.

Our target for this client is 12 revenue lines. At 12, it would be the most diversified property in the region — more diversified than the farm with the TV show.

Here's the part that matters for you: every fact in this issue came from public sources. Rate cards, booking pages, review platforms, ad archives, website code. No insider access. No paid databases.

Which means you can build this exact analysis for your own market. Below the paywall, I'll show you how we did it — where each piece of data lives, how to count revenue lines so you don't fool yourself, how to read a competitor's booking flow and marketing stack from the outside, and the three exploitable gaps we found hiding inside the region's #1 property.

How to build a competitive dossier from public sources

Everything above took two days. Here's the method, so you can run it in a weekend.

Step 1: Count revenue lines correctly

Read more

Your bookings didn’t drop. Your business was never built. If this summer’s numbers came in soft and you can’t say exactly why, one diagnostic call maps which problems are actually yours to fix. No pitch — you leave with the diagnosis either way.

Book your diagnostic →

Hey,

Got on a call a couple weeks ago with an operator running a two-bedroom in Oregon near a major national park.

Last summer he grossed around $8,000 a month, June through August. This summer, about half that.

His read on it: trail closures at the park killed his demand. He booked the call to figure out how to ride it out.

He wasn’t wrong about the closures. That part checked out.

He was wrong about what they explained.

I've spent 15+ years in this space, trained more than 10,000 operators through CashFlowDiary, and recorded 237+ podcast episodes breaking down the deals that work and the ones that don't. The pattern below shows up in every cycle.

• • •

Six Problems Were Hiding Behind One Excuse

By the end of the call, here’s what else was true.

Airbnb had rolled out a summer change to how it matches and ranks listings. His listing had taken a hit. He had no idea.

Airbnb’s Pro Hosting Tools had never been switched on.1 We turned them on live, on the call.

Every dollar of revenue came from the nightly rate. No add-ons. No upsells. Nothing underneath it.

He had never once captured a guest’s contact info. No repeat-booking system. No referral program. No newsletter.

And PriceLabs had been set one time and left alone.2

So yes, the trails closed. And six other things were quietly bleeding him at the same time.

The Drop Was a Diagnosis Arriving Late

The 50% drop wasn’t the problem. It was the diagnosis showing up late.

Every one of those problems existed last summer too. The strong market was hiding them.

When the calendar fills at $8,000 a month, nobody goes looking for the add-ons they’re not selling or the guest emails they’re not keeping. The number looks fine. The number was lying.

“A busy season pays the bill on a thin operation. It does not build the operation. Markets conceal weak businesses. They don’t fix them.” — J. Massey

The two-column audit: the storm outside vs the ledger inside.

How I Diagnose a Revenue Drop

First thing I do on a call like this: split the explanation into two piles.

Outside your control: trail closures, algorithm changes, the weather, the economy. That was never yours to fix.

Inside your control: everything else on his list. So we worked the inside part. That’s the only half that moves.

Then I count revenue lines. He had one — the nightly rate. One revenue line is fragile. When it dips, there’s nothing underneath it. Add-ons, upsells, mid-term stays, and early check-in fees are shock absorbers. When the nightly rate softens, they keep the floor from falling out. He had no floor.

Then I ask about guest data. He was paying Airbnb to send him strangers instead of building a list that books direct.

Last: tool drift. Platform tools don’t stay set. Algorithms change under you. Feature sets sit unused. A pricing tool set once and forgotten quietly taxes you every month you don’t touch it. His whole stack had drifted, and nobody was auditing it because the good season made the audit feel unnecessary.

One pillar vs four. Same house, different floor.

The list you own vs the audience you rent.

Run the Two-Column Audit This Month

You don’t need me for the first pass. Ask yourself: if your bookings fell 30% tomorrow, how much of the explanation would be inside your control? Then work this sequence.

1. Week 1 — Split the explanation. Two columns. Outside your control: weather, closures, the algorithm, the market. Inside it: revenue lines, guest data capture, repeat systems, pricing tools, unused platform features. Fill column two before you let yourself blame column one.

2. Week 2 — Audit the platform stack. Switch on every unused platform feature. Open the pricing tool and re-check its settings against the current season, not the season you set it in.

3. Week 3 — Start capturing guest data. Set up a simple, compliant way to collect contact info at booking and checkout. The list starts at zero. It only grows if it exists.

4. Week 4 — Add one revenue line. Early check-in fees, late checkout priced as a deterrent, a mid-term winter stay, one add-on guests keep asking for. One new line is enough to start building a floor.

Most of what shows up in column two was true last summer. You just couldn’t see it while the money was coming in.

Keep reading: the four levers that decide what your operation actually earns —

Common Questions About Booking Drops

Why did revenue drop 50% with the same property and the same season?

Trail closures reduced demand, but six structural problems were already running before the closures hit — an algorithm change, unactivated platform tools, a single revenue line, no guest data, no repeat systems, and a drifted pricing setup. They were bleeding revenue quietly during the strong season too.

What counts as a second revenue line for a short-term rental?

Add-ons, upsells, mid-term stays, and early check-in or late checkout fees all count. Each one is a shock absorber when the nightly rate softens. An operation with only a nightly rate has no floor when demand drops.

How often should operators audit their pricing tools and platform settings?

At minimum, once per quarter. Algorithm changes, new feature releases, and seasonal market shifts all move under your listing. A setup that was accurate in January may be working against you by June.

Ready to find out what’s actually costing you bookings?

One call found six fixable problems in this operator’s business. If your numbers dropped this summer and you can’t name which column they belong in, that’s exactly what the diagnostic is for.

Book your diagnostic →

• • •

P.S. The framework I ran on that call — Leads × Conversion × Frequency × Margin — is what decides what an operation actually earns. Most operators only ever touch the first lever. Next edition, I’m breaking down all four and the two nobody works.

You’re reading Cashflow Diary Direct — real numbers and working frameworks from an active STR operator, every week.

Ready for the next step?

Hey,

Today. 1:00 PM Central, 2:00 Eastern.

Three operators, one hour, every number real: the receipt, the live deal, and the banking math.

Your seat is free, and the replay goes to everyone who registers. The one more thing doesn't — that's for the room.

Register + join today's session →

See you at 1.

— J.