The 30-Call Owner Problem: Automated Owner Reporting That Gives GMs Their Week Back

Most hotels still build their reports by hand, and the general manager pays for it in calls, packs and Sunday-night spreadsheets. A reporting pipeline that pulls the systems overnight, reconciles them, drafts the variance commentary and logs every owner question can hand a GM most of that time back. Here is how it's built and what stays with the GM.

Look at the woman in the photo above. Phone to her ear, laptop open on the bed, two printed charts and her glasses next to it. Anyone who has run a hotel for an owner knows that call. It's Tuesday, the weekly pack went out Monday night, and somebody wants to know why rooms revenue is up and GOP isn't.

She'll answer it. Then she'll answer a version of it for the asset manager, and again on the brand call, and once more in the monthly narrative.

I owned ten hotels, so I've been the person on the other end of that phone. Later, as founder and CEO of Collective Retreats, I read reports from ten locations at once. Ten sites, a stack of systems at each, and the same handful of questions every week. Most of them could have been answered by the report if the report had been built to answer them.

That's the problem this piece is about. The fix is a reporting pipeline that does the assembling, reconciling and first-draft writing overnight, so the GM's job shrinks to checking, correcting and signing. I'll walk through what owners actually need at each cadence, what the published data says about the time it eats, how the pipeline works, what it saves, and the order to build it in.

The Problem: Role Creep, One Spreadsheet at a Time

Nobody hires a general manager to be a financial analyst. It happens anyway.

The management agreement says the operator will deliver monthly statements, a budget and a forecast. Stephen Rushmore's summary of what owners should demand lists monthly detailed P&L statements, cash flow forecasts and variance analyses, and those agreements often run 10, 20 or 30 years. Then the extras pile on. A daily flash. A weekly call. A pickup report for the asset manager. A brand call update. A lender covenant certificate once a quarter. Each one was reasonable when somebody asked for it. Together they eat the week.

Owners aren't wrong to ask. A survey of 499 general managers found that half of them describe owners as having moderate to strong influence over financial and operational decisions, and that owner involvement looked much the same at managed and independent hotels. Owners are involved because their money is at risk. When the information arrives late or in a form they can't use, they pick up the phone.

And the phone is expensive. The same GM is supposed to be walking the floor, coaching department heads and fixing the guest problem that's about to become a one-star review. The Axonify survey cited by HOTELS found 47 percent of front-line hospitality managers burned out and 64 percent had lost a team member to burnout. Reporting isn't the whole cause. It's the part a system can take.

Here's what a typical owner reporting stack looks like by cadence, and where each piece really comes from.

Look at the third column. Four to six systems feed every one of those packs. The GM's real job in reporting isn't analysis. It's copying numbers between systems and then explaining them.

The Data: Reporting Is Still Built by Hand

The most direct number comes from the distribution side of the business. The 2025 State of Distribution study from NYU's Tisch Center with RateGain and HEDNA, covering more than 700 hotel brands, found that 80 percent of hotels spend up to two days a week on manual reporting. Two days. That's a revenue and distribution team figure, not a GM figure, but the GM's weekly pack is built on top of those same hand-made reports.

Vendors selling the fix publish bigger claims, and they deserve a discount. Docyt describes daily financial reporting going from eight hours to thirty minutes of review, without giving the baseline behind the eight. Triptease describes a weekly revenue report at Lotte New York Palace that took 70-plus clicks per run and now runs automatically in minutes. Neither is a controlled study. Both describe the same mechanism, and the direction is right.

On the finance side, Gartner's 2025 survey of 183 CFOs found 59 percent of finance functions now use AI, up from 37 percent in 2023, but 91 percent reported low or moderate impact at the start. The top uses were knowledge management, accounts payable and anomaly detection. Reporting commentary wasn't near the top. That gap is the opportunity, and it's also a warning: buying the tool doesn't move the number.

Hotels show the same split. The AHLA and Actabl report released at The Hospitality Show this month found 84 percent of respondents say AI improved their own work, while 15 percent say it improved frontline work. Managers are getting the benefit first. Owner reporting is a manager's task, which is exactly why it's where the early gains land.

There's a second cost that doesn't show up in hours. LodgIQ's Samuel Johnson calls it decision lag: weekly packs circulate as proof of control, each department writes its own explanation, and by the time the owner reads it the week that mattered is gone.

The Framework: What a GM Should Still Own

Before building anything, sort the reporting work into three piles. I use this with every property, and it's the same sort we used in the month-end close piece.

Assembly is pulling numbers from systems into a format. It's all automatable, and none of it should touch a person.

Explanation is saying why a number moved. The first draft is automatable when the drivers are in the data (a group washed, a segment shifted, overtime ran in banquets). The final word isn't, because the GM knows things the data doesn't: the chef who quit, the road closure, the owner's son who comped his friends.

Judgment is what to do about it. Reforecast or hold. Cut hours or ride it out. Ask for capex or wait. That's the GM's job and the reason the owner hired an operator. A good pipeline exists to give the GM more time for this pile.

Here's what that sort looks like for one illustrative full-service GM's reporting week. It's a composite, not a measured property, built from the cadences in the first table.

Seven calls a week is about thirty a month. That's where the title comes from, and it's a conservative count for a GM with a hands-on owner. The model says about ten hours a week come back. Treat it as a planning number to test against your own logs, not a promise.

The Mechanism: A Reporting Pipeline in Five Parts

The pipeline isn't one product. It's five pieces, and most hotels already own some of them.

1. Nightly data pulls into one store. After night audit, connectors pull the PMS (occupancy, ADR, segment, on-the-books), POS (outlet revenue and covers), the timekeeping system (hours by department and job code), the revenue management system's forecast, and the comp-set feed. After month-end close, the accounting GL joins them. Everything lands in a single small database, keyed by date, department and USALI account. For one hotel this is a modest cloud database and a handful of API connectors; for a portfolio it's a proper warehouse. Where a system has no API, a scheduled export works fine. The point is that no human copies a number.

2. Reconciliation checks that run before anything is sent. The pipeline ties PMS room revenue to the GL, POS totals to the outlet lines, and timekeeping hours to payroll. It flags a break above a set tolerance instead of shipping it. This is the step most automated reporting skips, and it's why owners stop trusting the automated version after the first bad number. The check also compares today's figures to yesterday's for impossible jumps (a 400-room night on a 250-room hotel), which catches the duplicate import before the owner does.

3. Templated packs, built per reader. The flash, weekly and monthly packs are templates filled from the store. Each owner, asset manager and brand gets its own version, because they ask for different things and want them laid out their way. One build, many layouts. The templates should live in a shared place the GM can edit without calling a developer.

4. A variance commentary drafter. This is where the AI earns its place. For every line that misses budget or last year by more than a threshold (say 3 percent and $5,000), a language model reads the drivers in the store and drafts one or two plain sentences: rooms revenue was $42,000 under budget because a 180-room-night group washed to 110, partly offset by transient ADR $6 above forecast. Each sentence links to the numbers behind it. The model is told to say "driver not found in data" rather than guess, and the GM edits and approves every line before it goes out. Approve-before-send isn't optional. An owner who catches one invented explanation will never read the commentary again.

5. A forecast-to-actual log and an owner question log. Every forecast version is saved with its date, so the pipeline can score how close the hotel's forecasts ran (the method is in our forecast accuracy benchmarking piece). And every owner question, from the phone, email or text, goes into a log with the answer. An assistant drafts answers to new questions from the store, the GM approves them, and once a question shows up three times it becomes a permanent field in the pack.

That last part is the one that actually cuts calls. The report learns what the owner wants.

What the Owner Sees Change

The first thing an owner notices is timing. The flash arrives at 7 a.m. every day, not when the front office manager gets to it. The weekly pack arrives Monday morning with commentary already in it. The monthly report lands a few days after close instead of near the end of the following month, because the narrative isn't waiting on someone's weekend.

The second is consistency. Every pack uses the same definitions, so labor cost per occupied room means the same thing in week 12 as in week 40. Working across more than 800 hotels as VP of sales strategy and operations at Starwood taught me that a portfolio only reads cleanly when every property defines a number the same way. Half the arguments about whether a result is good turn out to be arguments about what the number means. A single store with fixed definitions ends most of those.

The third is the forecast record. An owner who can see that the hotel's 30-day forecast has landed within 2 percent for nine straight months gives the GM room to make calls. One who can't, calls.

That record has value well beyond the weekly call. Lenders and buyers ask for it too. A hotel that can show three years of system-generated forecasts against actuals walks into a refinancing or a sale with something most of its competitors can't produce.

Implementation: Twelve Weeks, In This Order

Weeks one and two are the baseline. Before you change anything, log every reporting task the GM, controller and revenue manager do for two weeks, with minutes, and log every owner, asset manager and brand question with who asked. Without a baseline you won't be able to show the owner what changed, and the owner is the one paying for the project.

Weeks three through five are the data pulls and reconciliation. Start with the PMS and the GL, because those two account for most of the owner's questions. Add POS and timekeeping next. Don't build a single pack until the reconciliation runs clean for a week.

Weeks six and seven, the daily flash and weekly pack go automated, running side by side with the manual versions. The GM compares them each morning. Any difference is either a pipeline bug or a manual error, and you'll find both.

Weeks eight through ten, turn on the commentary drafter for the weekly pack only. Set the threshold high at first so the GM sees a handful of drafts, not forty. Track how many drafts the GM keeps, edits or rewrites. If more than a third get rewritten, the driver data is thin, and that's a data fix, not a prompt fix.

Weeks eleven and twelve, the monthly report and the question log go live, and you send the owner a one-page note: what's automated, what the GM still writes, and the new delivery times. Ask the owner what they want added. It'll be the first entry in the question log.

Two warnings. Don't let the pipeline become one more system the GM has to log into; the packs should arrive by email or in the shared drive the owner already uses. And don't start with the annual budget. It's the biggest prize and the worst first project, because it happens once a year and you won't learn anything until next October.

If you want to know which of your systems can feed a pipeline like this today and which will need a workaround, that's the first thing our AI and technology scorecard and reporting engagement maps.

One more thing about the calls. Some of them should stay. An owner who calls because a number moved and wants to hear the GM's plan is doing the job right, and that's a conversation worth having. The pipeline's job is to make sure that's the only kind of call left.

Frequently Asked Questions