A message your automation sent is not an hour you worked, so it does not go in the log. Under Treas. Reg. §1.469-5T(f)(1), participation means work the individual actually performs in connection with the activity, which means the loggable part of a modern messaging setup is the work you do yourself: building the templates and rules, and handling everything the automation hands back to you. For the 2026 tax year, with scheduled messages, smart-lock codes and AI-drafted replies doing most of the routine sending, the honest hours column for guest communication is usually smaller than it was five years ago, and that is not a problem to be solved by rounding up.
What automation removes from the log
Automation is designed to take work away from you, and it succeeds. A platform that sends the check-in instructions at 9 a.m. on arrival day, the lock code an hour before, the mid-stay check-in on night two and the checkout reminder the night before has performed four tasks you used to perform. None of them is participation by you, because you were not working when they went out.
The same logic covers a smart lock that issues and expires codes on its own, a dynamic pricing tool that changes rates nightly, and an AI assistant that drafts a reply you send with one tap. The reply you read, edited and sent is a short piece of work. The draft that appeared without you is not.
Work performed by someone or something other than you is not your participation, and that is a point in favour of logging it honestly rather than against it. The earlier post on whether guest messages count as hours covers the manual case in detail. This one covers what is left once the robots have taken the routine sends.
Setup and rule-writing: real work, logged once
Configuring the automation is participation, and it is often substantial. Writing six message templates, deciding the triggers, testing the sequence on a dummy booking, connecting the lock to the calendar and fixing the two things that broke is a real afternoon of operational work on the activity you own.
Log it the way you would log any project: on the day you did it, with the task named and the minutes recorded. The distinction that matters is that it is logged once, not every time the system fires afterwards. Revisions count too. If you rewrite the check-in message in March because guests keep missing the parking instruction, that is twenty minutes in March.
Exceptions, escalations and problem guests: the loggable core
Once automation handles the routine, what remains in the guest-communication category is the exceptions, and exceptions are where the real minutes are:
- A guest replies to the automated night-two message about a problem, and you handle it.
- An inquiry asks something the templates do not cover, and you answer it yourself.
- A lock code fails and you talk a guest in at 11 p.m.
- A booking needs a date change, a refund discussion or a cancellation.
- A review needs a considered reply rather than a saved one.
This is a smaller set of entries than an unautomated host would have, and a better one. Each has a cause, a duration and a trace in the platform or your phone. Entries like these are what the court looked at in Mirch v. Commissioner (T.C. Memo 2025-128): the hours that survived were the ones with a message record behind them, and the categories logged in round blocks did not.
Why message counts are not minutes
The tempting shortcut, once a season's inbox is full of hundreds of threads, is to turn volume into time: so many messages, a few minutes each, call it thirty hours. Two problems with that.
First, most of those messages were sent by the system, so the arithmetic counts work nobody performed. Second, it is a formula, and formula-based reconstructions are exactly what the Tax Court has repeatedly described as estimates rather than a reasonable means of establishing participation. A log built from a multiplication is weaker than a shorter log built from events, even though the number at the bottom is bigger.
If you want a volume check, run it the other way: filter the inbox for threads you personally typed in, and see whether the count is in the same neighbourhood as your entries. A mismatch in either direction is worth understanding before year end rather than after.
Automation and the comparison prong
Automation has a second effect that owners rarely think about until a CPA raises it. The common STR test under §1.469-5T(a)(3) asks not only whether you participated for more than 100 hours but also whether you participated more than any other individual, and the 100-hour test guide walks through both halves.
Software is not an individual, so a scheduling tool does not compete with you in that comparison the way a co-host or a property manager does. But automation lowers your own total, which brings you closer to whoever is doing the physical work. A property where the cleaner logs 90 hours and your communication time has been automated down to 15 is a different picture than it was before you installed the tools. That is a reason to track the cleaner's hours and your own with equal care, not a reason to pad.
Honest estimates for short interactions
Many of these interactions are genuinely short, and short entries make owners uneasy because they look trivial. Log them anyway, at their real length. The what counts as participation guide explains why the character of the work matters more than its size.
Some practical conventions that keep short entries credible:
- Record the minutes you were working, not the span from the guest's first message to your last reply.
- Group one issue into one entry, even if it took four exchanges across an afternoon.
- Round to the nearest five minutes, in both directions.
- Do not log a glance at a notification. Reading a message and deciding it needs nothing is not a task.
A category that totals eighteen hours for the year and is traceable line by line is a stronger record than one that totals eighty and came from a multiplication.
Sample entries
| Date | Activity category | Entry | Duration |
|---|---|---|---|
| 02 Jun | Marketing & Listing | Wrote and tested six scheduled message templates, connected lock codes to calendar | 2h 10m |
| 14 Jun | Guest Communication | Guest reply to automated night-two message, broken blind, arranged handyman visit | 0h 35m |
| 21 Jun | Guest Communication | Lock code failed on arrival, phone call and manual code issued | 0h 15m |
| 28 Jun | Guest Communication | Date-change request, checked calendar, repriced, confirmed with guest | 0h 20m |
| 03 Jul | Marketing & Listing | Rewrote check-in template after three parking questions | 0h 20m |
Five entries, under four hours, and every one of them describes something a person did. Keeping setup work and exception handling in separate activity categories, which is how the entry form in STR Tracker is built, makes the pattern visible at year end: a block of configuration time early on, then a thin and steady line of exceptions through the season.
What to write down
For each entry: the date, the category, what prompted it, what you did, and the minutes. For the automation itself: a note of when you set it up and what it now sends, so that next year you can tell a CPA which parts of your inbox were worked and which were sent. If you want the whole category logged as it happens, from the phone the guest messaged you on, you can start a free trial.