Gross margin

Gross margin only means anything if the boundary is drawn consistently, and in agencies that boundary is drawn in several incompatible ways.

Above the line, in direct delivery cost, belongs anything that exists because this account exists: the salaried time of the people working on it, contractors and freelancers, and software licences billed specifically to the work. Below it, in overheads, belongs rent, leadership salaries, finance, central tooling and new business.

Pass-through media spend belongs in neither, because it is not revenue. An agency managing a hundred thousand a month of ad budget for a five thousand fee has five thousand of revenue on that account. Including the media makes the margin percentage look extraordinary and makes the agency price itself against a number that is mostly somebody else money passing through its bank account.

The test for the boundary: if this client left tomorrow, would this cost go away. Freelancer hours would. The office would not. That question resolves most of the ambiguous cases in a sentence.

An agency-wide gross margin is a number for a bank manager. The version that changes decisions is per client, and it is the calculation most agencies have never done.

The reason is mundane: it needs hours tracked against accounts and salaries allocated to those hours, and if either is missing the number cannot be produced. Which means the agencies that most need the insight are the ones least equipped to get it.

What it reveals is nearly always uncomfortable and nearly always actionable. A handful of accounts carry the business, a handful lose money, and the rest are unremarkable. The losses are rarely where anybody expected.

The pattern worth knowing in advance: the unprofitable account is usually the largest, the oldest, or the one belonging to a founder. Those are the three nobody questions, which is exactly why scope on them has drifted for years without anybody raising it.

An agency reporting a healthy blended margin can be carrying two accounts that lose money on every hour, and the blend will keep looking fine for as long as the profitable accounts hold.

That is not a healthy business with a rounding issue. It is a business with two problems and a statistic obscuring them, and the exposure is real: if one of the profitable accounts churns, the losses stop being covered and the picture changes in a single month.

The distribution matters more than the mean, and the practical version is a ranked list rather than a chart. Every client, sorted by margin, worst first. It takes one page and it is the most informative page in an agency management pack.

Reviewed quarterly, that list produces a short and specific agenda: reprice these two, renegotiate scope on this one, and understand why this one is better than we thought. None of those actions is available from a blended figure.

Once hours exist, they have to be costed, and the shortcut almost everybody takes is a blended hourly rate across the team. That shortcut hides the single most expensive pattern in agency delivery.

Senior people cost several times what junior people cost. Allocating everybody at an average makes an account staffed by the two most expensive people in the building look identical to one staffed by juniors, and it is frequently the cheap account that is soaking up the senior time, because it is the difficult one.

Costing at actual salary, even roughly banded, makes that visible immediately. The usual finding is that seniority has drifted toward the accounts that complain rather than the accounts that pay, which is a rational response by individuals and a bad outcome for the business.

And the last correlation worth naming: the worst three accounts by margin are almost always the three the delivery team finds most stressful. That is not coincidence, it is the same fact observed from two directions, and either one can be used to find the other.