In short
Revenue tells you what a client pays; it says nothing about what they cost. Estimate cost to serve by adding delivery hours to the unbilled surrounding work — meetings, revisions, chasing, admin and payment delay — which is where the difference between clients usually lives. A rough estimate across your top ten accounts is enough to change decisions. Expect to find that one or two large accounts are marginal and one or two small ones are excellent, and that the difference is behaviour rather than price.
Key takeaways
- Revenue per client is visible and nearly useless on its own. Cost to serve is where the variation is.
- The unbilled surround — meetings, revisions, chasing — is usually the difference between a good account and a bad one.
- Rough estimates across ten accounts beat precise tracking of one.
- An unprofitable client is usually a pricing or scoping failure, not a bad client.
- Payment behaviour is part of profitability. Ninety-day payers cost real money to finance.
Why revenue per client misleads
Every business knows its revenue by client, because that number falls out of the accounting system without effort. It is also the number most likely to lead you to the wrong conclusion, because it describes one side of a transaction.
Two clients each paying £40,000 a year can differ enormously. One approves work in a single round, has a decision-maker in the room, sends what you asked for, and pays in fourteen days. The other requires four calls a month, revises twice beyond the agreed scope, involves a stakeholder who joins late and changes direction, and pays at ninety days after two reminders. On the revenue line they are identical. On margin they are not remotely comparable, and the second one is quietly financed by the first.
The reason this persists is that all the difference sits in activity nobody records. Delivery hours might be tracked. The surrounding hours almost never are.
Nobody invoices for the fourth call, the third revision, or the two reminders. The client who needs all of them is not paying less; they are consuming more, and the consumption is invisible.
What cost to serve actually includes
A usable estimate has to include the surround, because the surround is where the variation lives.
| Component | Usually counted? | Varies between clients? |
|---|---|---|
| Delivery hours | Sometimes | Moderately |
| Meetings and calls | Rarely | Enormously |
| Revisions beyond the agreed number | Rarely | Enormously |
| Chasing inputs and approvals | Never | Enormously |
| Account management and reporting | Rarely | Substantially |
| Payment delay and chasing | Never | Substantially |
Notice the pattern: the components that vary most between clients are exactly the ones nobody counts. That is not a coincidence. Delivery is scoped and therefore counted; everything around it is absorbed as the cost of a relationship.
The last row is real money rather than an inconvenience. A client paying at ninety days on terms of thirty is using your working capital for two months, and if your delivery-to-invoice gap adds a further fortnight, as it does in most businesses that have not examined it, the financing cost belongs in this calculation. The mechanics of shortening that are in getting paid is an operations problem.
Estimating it without time tracking
Waiting for a time-tracking system means never doing this. A rough estimate across your top ten accounts takes an afternoon and is accurate enough to change decisions, because the differences are large.
- List the top ten by revenue. Beyond ten, the effort outweighs what you learn.
- For each, ask the delivery team for a monthly hours estimate, split into delivery and everything else. They will know. People who serve an account can tell you within twenty per cent, and twenty per cent does not change the ranking.
- Add account management and admin, including the person who chases the invoice.
- Apply one loaded hourly cost across the board. Salary plus overhead divided by realistic available hours, which is roughly 60–70% of contracted time, as set out in capacity planning.
- Compare to revenue and rank.
Accounts estimated roughly beats one account measured precisely. The purpose is a ranking, and a ranking survives a good deal of imprecision as long as the estimating method is applied consistently.
Do the estimate before looking at the answer, and do not adjust figures because the result is uncomfortable. The most valuable finding is usually the one that contradicts an assumption about a flagship account.
What the numbers usually show
Three patterns recur across service businesses.
The large account with thin margin. Big enough to matter, demanding enough to consume its own revenue. Frequently protected by the belief that its size makes it strategic, which is worth testing rather than assuming.
The small account with excellent margin. Clear briefs, few meetings, prompt payment. Usually invisible because it generates no drama, and usually the best template for the kind of client you should be seeking.
The account everyone dreads. Almost always the worst on margin. Dread is an unusually accurate proxy for unbilled effort, and it is available before any arithmetic.
The common thread is that the difference is behavioural, not commercial. Two clients on identical terms diverge because one has a decision-maker and the other has a committee, or because one sends the brief and the other sends fragments over three weeks. That means the lever is usually how the relationship is run rather than what it is priced at.
What to do about a poor account
The instinct on discovering an unprofitable client is to consider ending the relationship. That is a legitimate option and a poor first one, because most unprofitable accounts are mis-scoped rather than inherently bad.
| Cause | Response |
|---|---|
| Scope has quietly expanded | Reset the scope at renewal, using the record of what was added. This is why tracking additions matters even when you absorb them |
| Priced against an estimate that was wrong | Reprice at renewal using actuals rather than the original estimate |
| Client behaviour: meetings, indecision, late inputs | Change the process. Fewer, structured meetings; a named decision-maker; input deadlines with consequences |
| Your own delivery inefficiency | Not a client problem. Fixing it improves every account, not this one |
The fourth row is worth checking before any client conversation. If serving this account involves substantial rework or manual handling, the account is revealing an internal problem rather than causing one, and repricing would be charging the client for your friction.
Preventing the next one
Profitability analysis is retrospective by nature, but what it teaches is applicable at the point of sale.
- Price the surround, not just the delivery. If your best accounts consume six hours of meetings a month and your worst consume twenty, that difference belongs in the proposal as an included allowance.
- Write the input obligations into the agreement. What the client must provide and by when. Most overservicing starts with a client input arriving late.
- Name the decision-maker at kickoff. Accounts without one generate revision cycles indefinitely, and the cost is entirely yours.
- Re-run this annually. Accounts drift, and the drift is gradual enough that nobody notices until renewal.
The broader point is that profitability by client is an operations measurement wearing a finance label. What separates a good account from a poor one is almost never the rate; it is how much unpriced work surrounds the priced work. The Mayim Ops assessment measures that surrounding effort directly, which is why cost-to-serve problems usually appear in it as friction rather than as a pricing finding.
Frequently asked questions
How do you calculate client profitability?
Take the revenue from the client over a period and subtract an estimate of everything spent serving them: delivery hours at a loaded cost, plus meetings, revisions, account management, chasing and administration. Rough figures are sufficient, because the differences between clients are usually large enough to be visible through the imprecision.
What is cost to serve?
The full cost of delivering to a particular client, including the work that never appears on an invoice: calls, revisions beyond the agreed number, chasing information, extra reporting, and the financing cost of slow payment. It is normally where the variation between apparently similar clients is concentrated.
How do you know if a client is unprofitable?
Compare an estimate of hours consumed against revenue, and include the surrounding work rather than just delivery. A useful signal before doing any arithmetic is the client everyone quietly dreads: dread usually tracks unbilled effort closely.
What should you do with an unprofitable client?
Diagnose before acting. Most unprofitable accounts are mis-scoped or mis-priced rather than inherently bad, so the first options are repricing at renewal, tightening scope, or changing how the work is delivered. Ending the relationship is a reasonable last step, not a first one.
Do you need time tracking to measure client profitability?
No. An estimate assembled from the team's recollection across the top ten accounts is imprecise but unbiased enough to rank them, and ranking is what changes decisions. Precise tracking of one account tells you less than rough figures across all of them.