The most expensive project of my consulting life looked profitable until the day it ended. A $42,000 fixed-fee engagement, good client, work delivered on time, invoice paid without a murmur. Then our bookkeeper, new that quarter and annoyingly thorough, did something nobody had done before: she pulled every logged hour against it, added the hours people hadn’t logged but admitted to, and produced a number. After labor, the project had cleared about $3,800. A nine percent margin on work we’d priced to make forty. Nothing dramatic had gone wrong. That was the disturbing part. The money had left through six small holes, quietly, the way it almost always does.
That project taught me the thing this article is built on projects rarely lose money in an event, they lose it in a leak, and leaks have a geography. They happen in the same five places, in roughly the same order, on nearly every services project ever run, whether you’re an agency, a consultancy, a builder, or a dev shop. Learn the geography, watch a handful of numbers weekly instead of discovering them at the post-mortem, and you catch the leak while it’s still a decision instead of a result.
Where the Money Actually Leaves
The Estimate, Before Anyone Works a Minute:
The first leak happens before kickoff, in the proposal, and it’s the leak everyone downstream inherits. Estimates get built by optimists, usually the seller, usually anchored to what the client hoped to pay, and they consistently forget the same items: project management time, internal reviews, meetings, revisions, and the ramp-up hours where the team learns the client’s world. A useful discipline is brutal in its simplicity: pull the last five completed projects and compare estimated hours to actual hours, by phase. Almost every firm that does this for the first time finds a consistent gap, and the gap has a personality, always the same phases, often the same percentage. Your estimate error is not random, it’s a signature, and once you know yours, you price it in instead of donating it.
Scope Creep, the Famous One:
The leak everyone can name, and it’s rarely the biggest, but it’s the most preventable. Small asks accumulate one more revision, a quick extra page, a call that becomes a workshop. None of them individually justifies the awkwardness of a change order, which is exactly the mechanism, because scope creep is not a client behavior, it’s a pricing behavior, the sum of moments your team chose comfort over a conversation. The fix isn’t heroic backbone, it’s a lowered threshold: a standing rule that anything past a defined line, say two hours of unplanned work, triggers a written note to the client, even a friendly one, “happy to add this, it’s about X, want me to fold it in?” Most clients say yes. The ones who say no just saved you the hours.
The Gray Hours Nobody Logs:
Here’s the leak my $42,000 project bled from most, and the one time tracking data hides by design the work people do but never log. The quick Slack answer. The fifteen-minute “can you just look at this.” The internal debrief after the client call. Senior people are the worst offenders, both because their time costs the most and because they log the least, treating small assists as free. Individually the gray hours are rounding errors; across a project they routinely add 10 to 20 percent of true labor cost that appears in no report. You can’t fully capture them, but you can stop pretending they’re zero: apply a loading factor to logged hours when you assess margin, and make it culturally normal for a six-minute favor to get logged as one.
The Long Tail at the End:
Projects don’t end when the work is delivered. They end after the revisions, the walkthroughs, the “one small thing” requests, the handover documentation, and the six weeks of light-touch support nobody priced. The final 10 percent of a project regularly consumes a share of hours wildly out of proportion to its place on the timeline, and it’s pure margin, because the fee was recognized long before the tail stopped wagging. Watch for it in your data as the hours that keep landing on a project after its “end” date. If those tail hours are consistent across projects, they’re not an anomaly, they’re a line item your next proposal should contain.
The Gap Between Invoiced and Collected
The quietest leak isn’t in delivery at all. Work performed but not yet invoiced, and invoices sent but not yet paid, both age, and aging receivables on a project are margin evaporating in slow motion: financing costs, write-off risk, and the discount you’ll eventually accept just to close the file. A project isn’t profitable when the work is done or even when the invoice goes out. It’s profitable when the cash clears, and any accounting view that stops earlier is describing a hope.
The Numbers to Watch, Weekly
The whole point of project accounts done early is a short dashboard reviewed while the project is alive. Five numbers cover the five leaks, and none requires fancy software, just discipline:
| The number | What it catches | The early-warning sign |
|---|---|---|
| Budget burn vs. completion | Estimate error and creep | Hours 60% spent while work is 40% done |
| Effective hourly rate (fee ÷ all actual hours) | Every leak at once, in one figure | The rate drifting below your cost floor |
| Estimate vs. actual, by phase | Where your signature error lives | The same phase over-running on every project |
| Post-“end” hours | The long tail | Hours still landing weeks after delivery |
| Unbilled WIP and receivable aging | The cash gap | Work or invoices aging past 30 days |
The one to tattoo somewhere is the second. Effective hourly rate is the single most honest number in project accounting, because it can’t be argued with: take the fee, divide by every hour the project truly consumed, gray hours loaded in, and compare it against what an hour costs you fully burdened. Every leak on this list, whatever its source, eventually shows up as that one rate sinking. A project can feel busy, praised, and on schedule while its effective rate quietly drops through your cost floor, and that number is often the first place the truth appears, weeks before the P&L admits anything.
Making It a Habit Instead of a Post-Mortem
None of this requires a finance department. It requires a recurring 20 minutes, weekly, per active project, looking at those five numbers with the willingness to act on what they say, which usually means one of three small moves: a scope conversation with the client, a staffing adjustment, or a note to fix the next estimate. The firms that keep fat margins aren’t the ones with heroic projects. They’re the ones that catch the nine-percent project at week three, when it’s a correctable drift, instead of at the bookkeeper’s desk after the money’s gone.
We eventually re-priced the client from that $42,000 project, by the way, armed with the real hour counts, and the relationship survived it fine, because numbers make those conversations boring instead of tense. The new engagement made its margin. The old one earned its keep anyway, as tuition. Six leaks, one lesson, and a dashboard I’ve checked every week since.
