What your benchmark report actually measures
Every quarter a benchmark report lands in your inbox. You open it, look at page two, find out you're in the top quartile for something and the bottom half for something else, feel briefly good or briefly bad, and close it.
I did that for years before I read page three.
Page three is the methodology. It defines every number on page two. It is short, it is written plainly, and it contains four things that change what the report means. None of them are hidden — ServiceTitan publishes all of it, in their own words, in the same PDF. It is simply the page nobody opens.
Here is what's on it.
1. Your average ticket includes jobs that made no money
The definition, verbatim:
"Average Ticket = Total Job Average (average ticket across all completed jobs, including recalls, warranties and no-charge jobs)… this is the average for all jobs even if they didn't generate any revenue."
Read that twice, because it has a consequence most operators never connect.
Every callback you run lowers your average ticket. So does every warranty visit. So does every no-charge courtesy call. Those jobs enter the denominator with a value at or near zero and drag the average down.
Which means two companies with identical pricing can show materially different average tickets, purely because one has more rework than the other.
This matters because of what people do with the number. An owner sees a below-range average ticket and concludes it's a pricing problem. They rebuild the pricebook. They coach the technicians on presenting options. They push the comfort advisor on close rate.
And if the real cause was a callback rate running two or three points high, none of that helps — because the ticket wasn't measuring pricing in the first place. It was measuring pricing and quality, blended into one number, with no way to tell which was moving.
What to do: before you touch pricing because of a low average ticket, pull your true callback rate and your warranty volume. If either is high, recalculate your ticket across revenue-generating jobs only. That's the pricing number. The published one is a composite.
2. Booking rate has a sixty-second floor
"The Call Booking Rate is the % of call leads booked into jobs, where 'call lead' is an incoming call that lasts at least 60 seconds or is marked as a Lead."
This is a reasonable definition. It's also a choice, and it's not the one most owners have in their head.
Calls shorter than sixty seconds don't enter the denominator. Wrong numbers and misdials — correctly excluded. But also: the caller who hung up while ringing, the one who got a confusing greeting and bailed at forty seconds, the one who reached a voicemail they didn't want.
Some of your worst moments on the phone are exactly the shortest calls. Under this definition, they don't count against you.
This is not a criticism of the metric. You need some floor or every telemarketer counts as a missed booking. But it means your platform's booking rate and the number you'd get from "booked jobs ÷ every inbound call" are different metrics that happen to share a name.
What to do: compute both, once. Booked ÷ platform-defined call leads, and booked ÷ all inbound. The gap between them is your answer-rate and call-handling problem, and it's invisible in the benchmark number by design.
3. The comparison group is not the industry
"Only ServiceTitan customers who are Opted-In are used for Benchmarking."
So "higher than X% of Titans like you" means: higher than X% of ServiceTitan customers who opted into benchmarking, within your assigned peer group.
That's a genuinely useful comparison — those are real companies with real data, which is more than most industry benchmarks can say. But it is a specific population, and it skews in predictable ways: toward companies large enough to be on a full-featured platform, organised enough to have opted in, and by definition not including anyone running on paper, spreadsheets, or a competitor's software.
The peer grouping itself is built by "advanced analytics" on factors including business complexity, service mix, trade mix, business size and climate zone. The method isn't published.
What to do: keep using the percentile — just stop translating it to "the industry." It's a percentile of something specific, and knowing what changes how hard you should react to it.
4. The definition changed. It's disclosed in a footnote.
This is the one that does real damage, because it defeats the exact thing benchmark reports are used for.
Install and service jobs are classified algorithmically. The current rule uses a minimum ticket threshold — $100 for install, $50 for service. And then this, in small type:
"this definition was updated over prior versions to include a minimum threshold value. Previous versions had a threshold value of 0 for both install and service jobs."
The classifier changed. Mid-series. In a document people use to track themselves year over year.
I found this because the numbers stopped making sense. Across thirteen consecutive quarterly reports for one company, the peer group's install-to-service revenue ratio swung by more than thirty points between adjacent quarters — from roughly 42:58 in one to roughly 80:20 in another, and back.
No market does that. Not in a quarter. Not across an entire peer cohort simultaneously.
The market didn't move. The classifier did.
What to do: when a peer figure moves further in one period than any real market plausibly could, suspect the definition before you suspect the market. And when you're building a multi-year trend out of these reports — which is exactly what they invite — check whether the methodology page says anything changed. At some point, you are comparing two different measurements and calling the difference progress.
None of this makes the report useless
I want to be clear, because it would be easy to read this as an attack and it isn't one.
These reports contain something almost nothing else in this industry does: real operating data from real companies, compared consistently, delivered free, every quarter. The alternative for most operators is an association benchmark behind a membership paywall with an undisclosed sample, or a number someone repeated at a conference.
And ServiceTitan published every one of these definitions themselves. Nothing here was uncovered. It's on page three of a document they sent you. The failure is entirely in the reading.
The point isn't to distrust the report. It's to read it the way you'd read any other sourced claim: population first, definition second, number last.
Ten minutes, once
Open your most recent report and find the methodology page. Then answer four questions in writing:
- Who is in my comparison group? Write down the actual population, not "the industry."
- What's in the denominator of every rate I care about? Especially booking rate.
- What's bundled into the numerator? Especially average ticket.
- Has anything changed since last year's report? Compare the methodology pages, not the numbers.
Then write those four answers on the same page as your scorecard, permanently, so the next person who reads your dashboard inherits the definitions along with the numbers.
That's the entire discipline. A number without its definition isn't a measurement — it's a rumour with a decimal point.
If you want a second set of eyes on what your platform is actually telling you — and what it isn't — that's what the ServiceTitan Operating Audit is. One fixed-scope pass, one written baseline, no software to sell you.
This is IronMargin operational education, not legal, accounting, tax, or employment-law advice. See the disclaimer.
More from the blog