The Clients You Never Costed
A managing partner once quoted me his firm's realization to the decimal and, in the same breath, admitted he had no idea which quarter of his client list was underwater. That combination is normal. The firm-level number is easy to pull. The client-level number takes work nobody has been asked to do, so nobody does it.
Most accounting firms run on two figures. Total revenue, and the number at the bottom of the year. Both are firm-wide. They tell you whether the year was good. They say nothing about which clients made it good and which ones quietly ate the margin the good ones earned.
The average is hiding 2 clients you'd fire
Realization is the usual proxy for pricing health, and the benchmarks are clear enough. Firms that price well collect somewhere in the 90 to 95 percent range on standard work, according to the Cornerstone reporting on CPA billing and fees, and anything under about 85 percent is treated as a sign of underpricing. Fine. But a firm that reports 88 percent realization is not a firm where every client sits at 88. It's a book of 95s dragged down by a cluster of 70s, and the average tells you none of that.
Those 70s have names. One is the client whose scope crept 3 years ago and whose fee never followed. One sends every document in the second week of April, which turns a clean return into overtime and a write-down. Slow client document collection was the single biggest workflow problem firms named in 2025, and it hits realization directly, because the time you spend chasing paper is time you can't bill and won't recover. The firm-wide number absorbs all of it. The individual client, the one you would actually renegotiate or resign, disappears into a total.
The number you need is one you already track
Here's the part that should bother you. You are not missing the data. You track time. You just sum it the wrong way, up to the firm instead of down to the client.
Cost to serve a client is close to arithmetic. Take every hour worked on that account, not the hours you managed to bill, and multiply by a loaded cost per hour that includes salary, benefits, and the overhead each person carries. Set that against what the client actually paid you, collected, not invoiced. The gap is the client's real contribution. Do it for your top 20 accounts and I'll make a quiet bet. Two or three of them are running at a fraction of the margin you assumed, and at least one is costing you money to keep. You've been carrying it on the strength of the total.
This matters more in a firm than in most businesses because of a fact you already know in your gut. Accounting clients are sticky. They stay for years, they refer people like themselves, and the relationship is genuinely hard to replace. That stickiness is an asset when the client is profitable. When the client is underwater, it's the reason the loss compounds year after year while everyone assumes the account is fine because it's still there.
Run it on twenty, then decide
You don't need a costing system or a consultant for the first pass. You need a spreadsheet, your time records, and one afternoon.
Pull worked hours by client for your twenty largest accounts. Apply a loaded rate. Compare to cash collected. Rank the list by contribution margin, not by revenue, and look at the bottom five. Some of them you'll fix with a fee correction you should have made 2 years ago. Some you'll fix by tightening scope or by moving the work to a lower cost seat. And one or two you'll decide to let go, which frees the capacity to serve a client who pays you what the work is worth.
The firm-level number will never surface any of this, because it's designed to reassure you, not to inform you. It nets your best client against your worst and reports the blend as if the blend were a strategy. The clients hiding inside that average are the whole decision. Cost them individually, once, and you'll stop running the most important pricing call you make on a number that was built to hide it.