Agency project and retainer margin
Project margin is the fee for a piece of work minus every cost that work carried, at a fully loaded hourly rate. Calculated weekly and grouped by client and by service, it tells you which accounts earn and which ones the rest of the agency is subsidising. Calculated annually, it tells you nothing you can still act on.
Written for founder-led UK agencies of roughly £500k–£2M. An agency can hold a respectable overall margin while a third of its accounts lose money.
The four numbers, per account
Four figures per live client, updated weekly by the person running the account. Fee, cost to date at a loaded rate, forecast to complete, and work delivered beyond the agreement. The fourth is the one nobody records and the one that explains most of the gap.
| Number | What it includes | Where agencies get it wrong |
|---|---|---|
| Fee | Invoiced value for the period, plus priced extras the client has agreed. | Counting verbally agreed extras nobody has priced or raised. |
| Cost to date | Hours at a loaded rate, plus freelancers, media, licences and anything bought for the account. | Costing hours at salary rather than salary plus employer costs plus overhead share. |
| Forecast to complete | What the account lead believes the rest of the month or project will take. | Leaving it out, which removes the only figure you can still act on. |
| Unbilled scope | Work delivered outside the agreement, in pounds, running total. | Calling it relationship investment and never totalling it. |
Why retainers hide the worst of it
A retainer conceals margin decay because the fee stays flat while the work grows. A retainer priced for four workstreams that now delivers seven looks stable on the revenue line and loses money on the cost line. Everyone is busy, the client is happy, and the account slid two years ago.
Three things drive the slide, and all three are gradual enough that no single week feels like the moment to raise it.
The client's team changed and the scope did not. A new marketing lead asks for things the last one did not. Your team absorbs them to build the relationship.
The agency added value to keep the account. Extra reporting, an additional review call, a monthly session that was never in the fee. Each one is small and permanent.
The fee never moved. Salaries rose, the work grew, and the retainer renewed at last year's number because raising it felt risky in a quarter that was already tight.
Review every retainer against delivered hours twice a year. Agencies that do this find one or two accounts that need re-pricing or ending, and the conversation is easier than the founder expects.
Three decisions the number should trigger
Re-price or release
An account below your margin floor for two consecutive quarters gets a new fee or a notice period. Set the floor in advance so the decision is arithmetic rather than emotion.
Change who delivers it
If seniors are delivering work priced at mid-weight rates, either the price or the resourcing is wrong. Both are fixable in a month.
Price the scope creep
Total the unbilled work, take it to the client as one conversation, and set a written limit on what an account lead may absorb without asking.
The mechanism and the weekly meeting that runs on it are set out in project margin that arrives too late. Utilisation tells you whether people are busy; this tells you whether being busy was worth it.
What the correction looks like
One UK digital agency, growing its revenue and losing money, where margin stayed invisible until the year closed.
Common questions
What margin should an agency project make?
Set the floor from your own overhead and the profit you intend to keep, then hold every account against it. The useful comparison is internal: your best-performing service line against your worst, and this quarter against last. An external average built on a different salary bill and rate card cannot tell you whether a specific account is worth keeping.
How do we cost an hour properly?
Take salary plus employer costs plus the tools that person needs, divide by their available hours, then add a share of overhead. Costing at salary alone understates the true cost by a wide margin and makes loss-making work look profitable, which is how agencies keep selling the service that is quietly funding itself from the others.
Should we drop a client that is losing money?
Re-price first. Most underwater accounts were priced correctly at the start and grew without the fee moving, so a scope conversation and a new number solves it. Release the account when the client declines the new price and the work cannot be reduced to fit the old one.
How often should we look at this?
Weekly for live projects and month-to-date retainers, because that is the only frequency at which you can still change scope, price or staffing. Quarterly for the pattern by service line, which tells you what to stop selling. Annually you are reading history.
Related reading: Fractional COO for agencies · Agency utilisation rate · Agency WIP and payment terms · Busier than ever, and the bank balance disagrees
Find the account carrying the rest
Bring three accounts and a rough sense of the hours each one takes. That is usually enough to show where the fee stopped matching the work.
Book an Operational Clarity Call →