🏠 GEM Resource Hub · 📚 Library · 📘 About RocketLevel · 📘 Playbooks · 🗺️ What's Your Market Worth? · 📅 Book a Call

How each operating number is computed, where it comes from, who takes it, and where to write it down, so next month's reading has something to sit beside.

A guide for restoration owners and whoever closes the books, backed by decades of experience in the trade. Eight numbers, taken the same way every month, by a named person, into one sheet that keeps its own history. About a 9-minute read.

<aside> 💡

The principle: a buyer at exit, a banker underwriting a line, and a surety setting your bond capacity all open the same three documents first: your work-in-progress (WIP) schedule, your aged receivables, and three years of net. If you can hand those over on a Tuesday afternoon without a scramble, the number they write down is far more likely to be the number you earned.

</aside>

<aside> 🧭

Which one do I need? This is the financial operating scorecard: EBITDA (earnings before interest, taxes, depreciation and amortization), DSO (days sales outstanding), WIP, margin, the numbers you run the business on. For the sales-activity scoreboard, dials, meetings, show rate, pipeline coverage, the numbers you run reps on, see The Commercial Sales Scoreboard: The Numbers You Run Reps On. To put either on a weekly cadence, see Run Your Commercial Business on an Operating Rhythm.

</aside>

• 📑 Jump to a section

Why there is no benchmark column on this page

We left a healthy range off every metric on purpose.

The published benchmarks move every edition and split hard by revenue band. Take overhead as a share of revenue. Editions of the RIA Cost of Doing Business Report report materially different figures year to year, partly because the industry moved and partly because a different set of companies answered, and the ratio moves hard with company size, so an industry-wide average is close to useless as a target. The 2025 edition, which reports FY2024 data, puts average total overhead at 38.53% of revenue on page 37, and the revenue-band spread on pages 38 and 39 runs from 26.55% to 45.97%. Use your band. Net margin is worse. Trade coverage has quoted different figures, so we went to the source: the 2025 edition puts nearly half of companies (48%) below 10% net income in FY2024, down from 52% in FY2023 (page 23). Its distribution on page 22 splits that into 41% with a positive net under 10% and 7% negative, with 23% at 20% or better. Trade coverage that quoted 52% was quoting the prior year. We are not going to pick one of those and print it next to your number as though it settled something.

If you want the outside figure for your own band, get it properly rather than secondhand. The 2025 edition is a free download from the Restoration Industry Association (checked September 25, 2026). When it arrives, read your own revenue band rather than the industry line, and write the edition year beside every figure you copy out.

Record: Benchmark source we actually hold: · Edition year: · Revenue band we read: · Date obtained: · If we hold none, write None:

Your own baseline is the benchmark

The useful comparison is against your own shop last month.

Take the reading this month. Take it again next month, computed the same way, by the same person, from the same report. Then read the direction rather than the level. A single reading carries whatever happened to land in that month: one large loss, one carrier paying late, one truck bought outright. Direction across several months is what strips that out.

How many months before a direction means anything

Four readings. That is a RocketLevel working threshold, and the arithmetic under it is below so you can argue with it.

Four monthly readings give three moves between them. If a metric were pure noise, each move is up or down with equal chance, which gives eight equally likely sequences of three moves. Two of those eight run all one way. So three moves in the same direction turns up by chance about one time in four. Five readings give four moves and sixteen sequences, two of which run one way, which is one time in eight.

Now check it from the other end, because that is the number that should worry you. This scorecard carries eight metrics. Eight metrics at one in four is two spurious directions on the board every month. Eight metrics at one in eight is one. That is exactly why the bands below let a three-move run ask a question rather than authorize a spend, and why even a four-move run gets confirmed against a second metric that ought to be moving with it.

What the threshold proxies for. Four consecutive readings moving one way stands in for a change in how the business is being run, as against a change in what happened to land that month. You cannot measure the first thing directly. Persistence is the cheapest honest proxy for it.