Client profitability: the number agencies calculate wrong

The arithmetic is worth writing out, because the omission is easier to see in the formula than in the result. Direct client profitability takes revenue billed to that client over a period, subtracts the salary and loaded cost of the people who did the work, and sometimes subtracts pass-through third-party costs like media spend or a subcontractor. What is left is the client profit. It is a real and useful measure of delivery efficiency, and it is routinely presented as though it answered a different and larger question: should this client be on our list next year?

Those are different questions, and only one of them is being calculated. Delivery efficiency is a good thing to know. Whether a relationship is worth having requires knowing what it cost to acquire, to run, and to collect from — and none of those appear in the formula, because each one lives outside the project, outside the timesheet your account manager fills in, and outside the invoice the client paid.

All figures below are invented, for illustration. This is not a real client, a real agency, or a real quarter, and nothing here reports how your business actually performs. What it shows is the shape of the arithmetic.

Take the agency's most-praised account: a large, reasonably prompt, high-referral client whose work is genuinely good. In a quarter they are billed ₹9,00,000. The delivery effort behind that is 540 hours at the agency's fully loaded rate of ₹1,450, so ₹7,83,000 of direct cost, plus a third-party cost of nothing because there is none. Direct client profitability is ₹1,17,000 — 13.0%. Nobody is alarmed. The client is on the retention list and the account manager describes them as low-maintenance.

Now count the effort that sits outside the project and never reaches an invoice. The team spent 46 hours quoting and re-quoting for this client, including a re-quote after the scope moved. Someone spent 32 hours on account management: a weekly call, two reviews, a quarterly deck. Someone spent 21 hours chasing payment, because this client pays at roughly 78 days against 30-day terms and each invoice needs several reminders. And 64 hours went into re-work on a module whose acceptance criteria were never written down, mostly senior time, mostly good people doing careful work on an ambiguity nobody had closed.

The fully allocated picture is now straightforward arithmetic. Direct cost was ₹7,83,000; the unallocated effort adds ₹2,36,350, giving ₹10,19,350 against ₹9,00,000 of revenue. This client is ₹1,19,350 a quarter in the hole, and the firm has been describing them as a model account. Note the total hours as well: 703. That works out at roughly ₹1,280 of revenue for every hour of the agency's attention, against a loaded cost of ₹1,450. On a fully allocated basis, every hour this client consumes is sold below cost.

Two of these deserve more than a line. Re-work is the most dangerous, because it grows with revenue rather than shrinking with it: a client can be paying you more and consuming more, and the extra effort appears as normal delivery rather than as loss. The tell is that your senior people are spending their time on work the client already paid to have done once. Late payment is the second, and it is the one most often misfiled as a collections problem rather than a profitability one — the money eventually arrives, so nothing looks broken, while the effort required to make it arrive is a permanent tax on the relationship paid in your team's hours.

Small engagements, for the same reason as in project economics: the measurement costs more than the error it catches. A one-off piece of work with a short selling cycle and a single invoice does not need an allocated effort model, and a firm that spends its afternoon building one instead of doing the work has made a bad trade. A practical threshold is whether the effort involved in acquiring, running, and collecting from that client is large enough to be worth a conversation when it grows.

There is also an honest counter-argument, and it deserves stating. Some clients pay a premium precisely because they are low-effort, and some high-effort clients are strategically worth subsidising because they refer others, carry your name, or hold a category open. That is a legitimate business decision. The objection is only to making it by accident — to never computing the number and therefore never knowing that you made it.

The discipline this imposes is the same one we hold to everywhere else in the product: an allocation must be visible enough to be argued with. The moment it is a column in a spreadsheet somebody built, it can be checked, challenged, and changed deliberately. The moment it is absent, the number is not conservative — it is simply unknown, and unknown is what gets mistaken for healthy.

We want to be direct about something that is often sold as a calculation and is not one. There is no neutral answer to how much of a shared account manager's week belongs to the client who calls most. Two owners in the same industry, with the same records, will allocate a pre-sales hour differently — one will charge it to the opportunity, the other to overhead — and both are defensible. The same is true of whether to absorb your office rent into a client figure or report contribution instead. These are choices, and choosing well depends on what you are trying to decide.

The argument for activity-based costing is instructive here, including its well-documented weakness: it assigns overhead to activities rather than spreading it on a single volume basis, precisely because the choice of driver is what determines whether the result is meaningful, and choosing a wrong driver produces a number that is confidently wrong. That is the shape of the whole problem. There is no driver that makes the allocation automatically correct, because the allocation encodes a view about your own business that only you hold.

So the point is not that one allocation is right. The point is that the allocation must be deliberate and visible rather than absent. An agency that writes down its four categories, applies one rate consistently, and looks at the result every quarter can change its mind about the answer. An agency that never allocates cannot change its mind, because it has never stated what it believed in the first place — and that is the failure, not the judgement itself.