What is Gross Margin per Engagement?

A people business can be profitable in aggregate while individual engagements quietly drain it, and the aggregate number will never say which. The annual margin is announced, the portfolio is declared healthy, and somewhere inside the average a handful of engagements are consuming senior attention, absorbing scope and losing money on every hour, subsidised by engagements that nobody is studying either.

The honest unit of profitability

Gross margin per engagement is revenue minus direct delivery cost, measured for one engagement at a time. The revenue side is what was actually recognised on the work. The cost side is what it actually took to deliver: the hours worked, priced at the loaded cost of the people who worked them, plus direct expenses. The engagement is the honest unit because it is the unit of commitment: one promise, one price, one team, one statement of work. Every profitability question a firm cares about eventually resolves to this grain, because this is the grain at which someone decided something.

The arithmetic is simple, and the numbers here are invented purely to illustrate it. An engagement recognises one hundred and twenty thousand of revenue in a quarter. The hours worked on it, at the loaded cost of the actual people involved, come to seventy eight thousand, and direct expenses add six thousand. Gross margin is thirty six thousand, or thirty percent. The interesting questions all start there: is that before or after the delivery lead’s unrecorded oversight time, and how does it compare with the margin the deal was sold at?

The allocation traps

The number is honest only if the cost line is, and three traps corrupt it. The first is over-allocation: spreading office, tooling and corporate overhead across engagements by formula until every engagement looks marginal. The number stops being believed, and meetings dissolve into arguments about the allocation instead of decisions about the work. The second is the opposite and quieter trap, under-counting: management and oversight time that was never recorded against the engagement, senior firefighting booked to a general code, rework absorbed in evenings, and bench time held for the engagement between phases. These are real, direct consequences of the engagement existing, and leaving them out flatters exactly the engagements that need scrutiny most. The third is spreading shared costs by revenue, which taxes the largest engagements for being large and excuses small, demanding ones. The working principle: a consistent, defensible direct-cost line beats a theoretically complete one, because consistency is what lets you compare engagements and trust the ranking.

What the portfolio average hides

An average is a statement about a distribution that refuses to describe it. A comfortable portfolio margin is compatible with a tight cluster of similar engagements, and equally compatible with a barbell: a strong head funding a draining tail. The two portfolios deserve completely different decisions, and the average cannot tell them apart. Portfolio profitability is not a fact about the portfolio. It is the sum of engagement-level facts that the average has been hiding. The tail engagements are rarely just unprofitable; they are usually also the loudest, the heaviest consumers of senior attention and goodwill, which is cost-to-serve compounding the margin problem. And the average hides motion as well as position: an engagement drifting steadily downward can be offset, in the blend, by a new one starting high, so the aggregate holds flat while a specific promise deteriorates in plain sight.

One concrete example

Clearly illustrative, with invented round numbers and no customer implied. A firm of a few hundred people runs about twenty active engagements and reports margin at practice level, where every practice looks acceptable. Someone finally builds the engagement-level view. The ranking shows a short head of engagements earning most of the profit, a broad unremarkable middle, and four engagements at the bottom that turn negative once the delivery directors’ firefighting time is counted against them. Two of the four belong to the firm’s proudest logo. Nobody had seen the ranking before, because no report had ever been produced at the grain where the losses lived, and the practice averages had absorbed them for years. The book chapter The Margin Leak follows this anatomy further.

The decision the number should trigger

A ranked engagement margin view is not reporting; it is a queue of decisions. For each engagement in the tail there are four honest options: reprice it, rescope it, restaff it, or exit it, each with a cost and a consequence that can be estimated and compared. For the head, the decision is to understand it: what those engagements share in shape, client, pricing and team, and how that pattern should change what the firm sells next. The decision-intelligence discipline turns the review into a record: options priced, the choice and the chooser captured, the outcome scored later against what was expected. Done that way, the firm learns which rescues work and which drains never recover, and the portfolio average goes back to being what it should have been all along: a summary of engagements the firm actually understands, not a blanket over the ones it does not.