The best quartile spends 43% of revenue on labor. The worst spends 68%.

Sixty percent of the day on admin. Twenty five hours a week on repetitive tasks. Eight to twelve hours on tenant messages. Those three figures turn up in almost everything written about property management labor.

We spent a week trying to trace them. All three dead-end at a vendor blog citing an industry survey nobody names, linking to nothing. Follow them far enough and they just run out.

What is left after you throw those out is smaller, duller and a great deal more useful. One number in this industry is measured properly, and the spread inside it is 25 points wide. You can work out which side of that spread you are on this afternoon, from a statement you already have.

Labor as a share of revenue is the number that is measured properly

NARPM publishes a benchmark study of its members' books, produced with ProfitCoach. The 2022 Financial Performance Guide covers 2021 financials from 153 companies. The financial data came out of QuickBooks and their property management software rather than off a survey form. It was then reviewed four times against NARPM's accounting standards.

That is the figure people quote. It is also the least interesting one in the study, because almost nobody sits on the average.

The spread is 25 points wide, and that is the whole story

Split the same 153 companies into quartiles by profitability and the labor line moves a long way:

• Least profitable quarter: 68.5% of revenue went to labor
• Second quarter: 57.9%
• Third quarter: 52.9%
• Most profitable quarter: 43.1%

Twenty five points, on one line item, in one industry, in one year. Two companies billing the same revenue can be spending a quarter of it differently on people.

The gap between the best quartile and the worst is 25 points of revenue. Nothing else anyone tells you about your labor costs matters until you know where you fall.

It takes one division on your own profit and loss statement. It costs nothing, and nobody's got to sell you anything first.

Doors per person: 49.3 on average, 58.6 in the top quarter

The same study reports units per direct team member. Three caveats sit on that figure, and the study states all three plainly:

• It excludes companies where more than 20% of occupied units were multifamily. These are single family numbers.
• Headcount was reported by the companies themselves. The financials came out of their accounting systems, so this is the softer half of the study.
• The data is from 2021.

You'll see other doors-per-person figures quoted constantly. Forty doors per hire is common. So is fifty. We could not trace either one to a primary source. AppFolio did publish a figure of 54 units per employee in 2025, and it is worth knowing what that is: AppFolio calculated it by dividing Census Bureau rental unit estimates by Bureau of Labor Statistics employment data. It doesn't come from their platform. AppFolio also notes, to their credit, that the real workload is higher, because the BLS figure counts admin, accounting and sales staff as well as managers.

Maintenance runs about 3.4 requests per unit per year

Maintenance is the engine underneath all of this. Hemlane published an analysis of more than 193,000 maintenance requests from its own platform covering 2018 to 2026. Single family rentals generated 3.40 requests per unit per year. Multifamily generated 3.23.

This is vendor platform data from a self selected customer base, so treat it as a starting point rather than a law. But it gives you something concrete to check your own system against. At 3.4 requests per unit per year, a 300 door portfolio produces roughly 1,020 maintenance requests annually. That is about four every working day, before anybody calls about anything else.

The category split is worth knowing too: plumbing 25.4%, HVAC 12.3%, appliance repair 12.2%. Roughly half of everything that comes in is one of those three.

Six in ten companies in this industry are five people or fewer

Buildium surveyed 1,796 property management professionals in June 2024 for its 2025 industry report. Sixteen percent were a single person. Another 46% had between two and five people, including the person answering. And 44% managed 100 units or fewer.

Keep that in mind when you read anything written about property operations, including this one. Advice built around a dedicated leasing team and a controller doesn't describe the median company. In most of them, the person doing the admin is the owner, or one of the two people they trust with it.

Clerical work is the only job group rated highly exposed to AI

Here the research is better than you would expect, and a good deal more modest than the marketing.

The International Labour Organization published a task level study of generative AI exposure in 2023. Rather than guessing at whole jobs, it scored the individual tasks inside each occupation. Clerical work came out as the only broad occupational group that was highly exposed: 24% of clerical tasks were rated highly exposed, and 82% were exposed at an above average level. Every other occupational group ran between 1% and 4%.

Read that carefully, because it cuts both ways. The admin layer of a property management company genuinely is among the most exposed work in the economy. And the ILO's own conclusion was augmentation rather than replacement. What they found is that the tasks inside a role change, not that the role goes away. It's a narrower claim than the headlines made of it.

The other paper you will see quoted, Frey and Osborne's 2013 estimate that 47% of US employment sat in a high risk category, was written a decade before large language models existed and never was a prediction of job losses, so treat it as history rather than evidence.

Nobody has measured the after-hours call, or the week

There is no credible study of how a property manager's week divides into hours. We looked hard. Nothing exists with a stated method and a sample size behind it.

We found nothing usable on after hours call volume either, or on what share of after hours calls turn out to be real emergencies, or on how many calls a portfolio of a given size generates in a month. Every figure in circulation on those questions traces back to an answering service's marketing page.

The strongest labor finding in the data is not software

If you came here expecting software to be the answer, the NARPM data has a complication in it worth sitting with. The lever most visible in those books is offshore staffing.

Companies in the study used global team members for 32% of their labor on average, and 65% in the top quartile. Average 2021 profitability rose from 9% at companies with none to 16% at companies with six or more. The study found no matching drop in churn or review scores.

We sell software and that is still the strongest labor finding in the data. Any honest account of this industry has to say so.

It also points at something more useful than a fight between the two. Offshore staffing and good software are answers to the same question: how much of this work has to be done by an expensive person sitting in your office. Ask that question properly and you can answer it with people, with software, or with both.

Three numbers to produce this week

None of these need a consultant, and all three come out of systems you already own.

1. Your labor line as a percentage of revenue. Compare it to 43.1% and 68.5%, not to the 55.6% average.
2. Your occupied doors divided by your direct staff. Compare it to 49.3, remembering that figure is single family, self reported, and from 2021.
3. Your maintenance requests per unit last year. Compare it to 3.4.

If all three land near the good end, labor efficiency is probably not your problem, and no software's going to invent one for you. If they land near the bad end, you now have a rough size for the gap, in your own numbers, from your own books.

Which is a better place to start than the three figures at the top of this page, and everything else built on top of them.