A client of mine had the best quarter in the agency’s history. Revenue up, margin up, utilization exactly where you would want it. Six weeks later their largest account left, and two of the three people who ran it left with them.
Nothing in the reporting was wrong. Every number was accurate, closed on time, properly coded. The reporting simply had no way of knowing that the account lead had been apologising to that client for four months, or that the reason utilization looked so healthy was that the same three people had been absorbing scope creep rather than escalating it.
A number is a summary, and summaries drop things
This is not a criticism of accounting. It is what accounting is for. A financial statement takes several thousand events and compresses them into a page you can hold in your head, and that compression is the entire value of the exercise. Nobody can run a business by reading the general ledger line by line.
But compression is lossy. Everything that made it onto the page did so because somebody, at some point, decided it was the kind of thing worth counting. Everything else is not merely unreported. It is invisible, and invisible in a specific and dangerous way: the report does not have a blank space where it sits. The page looks complete.
So the useful question about any report is not just what does this say. It is what would this look like if the thing I am worried about were happening. If the answer is “exactly the same,” you have found the edge of what the numbers can do for you.
Four things an agency P&L cannot see
These are the four that have cost the agencies I have worked with the most money, and none of them appear anywhere in a standard monthly package.
The health of the client relationship
Revenue is a trailing record of a decision the client already made. It tells you nothing about the decision they are about to make. An account can bill at full retainer for eight straight months while the relationship quietly ends, because notice periods and inertia keep the invoices flowing long after the goodwill has gone.
Whether your people are about to leave
Payroll is your largest line item and it reports headcount, not commitment. The month before a key person resigns looks identical to the month before they do not. In a business where the assets go home every evening, that is a significant amount of the enterprise value sitting outside the reporting entirely.
How the work was actually produced
A project delivered on budget by a well-run team and a project delivered on budget by three people working weekends land on the P&L in exactly the same place. One of those is repeatable. The margin looks the same either way, right up until the point it very much does not.
The quality of what you shipped
There is no line for whether the work was any good. For a business whose growth depends almost entirely on referral and reputation, the single largest driver of next year’s revenue is the thing this year’s statements are least able to measure. This is true across creative and communications shops alike, and it is why two agencies with identical financials can have completely different futures.
What gets measured gets managed, and then gets gamed
The standard response to all of this is to measure more things, and up to a point that works. Utilization, AGI, WIP, and client-level profitability are genuinely useful, and an agency that does not track them is guessing. I would not want to run one without them.
The trouble starts when a measure becomes a target. Put a utilization number on a wall and it will go up, because people are resourceful and there are a dozen ways to make that particular number move that have nothing to do with the business getting better. Time gets recoded. Work that should have triggered a scope conversation gets absorbed. Nobody is being dishonest. They are responding rationally to what the organisation has told them it cares about.
Which means every metric you elevate quietly reshapes behaviour around itself, and the more weight you put on it the less it tells you. The measure was a proxy for something you actually wanted, and proxies degrade under pressure. A good CFO holds two things at once here: take the number seriously, and never entirely trust it.
A model can only see what somebody wrote down
We build a lot of automation for agency finance teams, so I want to be careful here. The tooling is good and getting better, and I am not making a romantic argument for intuition over data.
But a model trained on your financial history has access to precisely the slice that got recorded, and it will produce a confident, well-formatted answer from that slice without ever indicating what was missing from it. Human judgement at least has the decency to feel uncertain. The output of a system that cannot see the account lead’s worried expression looks exactly as authoritative as the output of one that can, and it arrives faster, in better English, with a chart.
That is why the systems we build are designed to surface questions rather than settle them, and why a person stays in the loop as a verifier rather than a cleanup crew. Automation is superb at the compression. It is the decompression, the part where you work out what a moving number actually means, that still requires somebody who has spoken to the client this month.
Five questions that live outside the report
These take one meeting a month and they consistently surface things the close never will. Ask them of the account leads, not the finance team.
Which client would be hardest to replace, and how are they feeling?
Concentration risk is a number. How that concentration is holding up is not. Find out before the notice period does it for you.
Who on this team would be hardest to lose?
Then ask when anyone last spoke to them about anything other than deadlines. This is a finance question, whatever it sounds like.
What did we deliver this month that we are proud of?
If nobody can name anything, the referral pipeline is already thinning and the revenue will follow in about four quarters.
What are we absorbing that we have not billed?
Unpriced scope is real cost hiding inside a healthy-looking margin. It is also the clearest early signal that a pricing structure has stopped working.
What would have to be true for this number to be misleading?
Ask it of whichever figure you are most pleased with. That is the one nobody is checking.
None of these produce data you can chart, and that is rather the point. They produce the context that makes the chart mean something, and they do it while there is still time to act. A wider set of these questions sits behind how we run the work.
Common questions
What are the limits of financial data?
A financial report is a compressed summary of a period that has already ended. It records what was transacted, not why, and it cannot see the things nobody thought to count: goodwill with a client, whether your best team is about to leave, how much of last quarter ran on unpaid heroics. The numbers are a reliable record of a narrow slice, and the slice is narrower than most people assume.
Why do good numbers sometimes precede a bad quarter?
Because most financial reporting is lagging. Client relationships, team morale, and quality of delivery erode for months before any of it shows up in revenue or margin. By the time the P&L reflects the problem, the problem is old. That gap between the leading reality and the lagging report is where most unpleasant surprises live.
Should agencies stop tracking metrics like utilization?
No. Utilization, AGI, and WIP are genuinely useful and an agency that ignores them is flying blind. The discipline is remembering what each one measures and what it does not. Utilization tells you how much time was billed. It says nothing about whether the work was good, whether the client noticed, or whether the person doing it is on their way out the door.
How does a CFO use information that isn’t in the numbers?
By treating the report as the beginning of the enquiry rather than the end of it. A number that moves is a question, not an answer. The job is to go find the cause, which almost always lives in a conversation with the account lead, the delivery team, or the client, and then decide what to do while there is still time to do it.
The bottom line
Reporting earns its keep by telling you the truth about a narrow slice of the business, quickly and reliably. That is a genuinely hard thing to do well and most agencies do not have it. Get it right first.
Then hold it lightly. The number tells you something changed. It does not tell you why, and the why is where every decision worth making actually happens. That is the job we think a CFO is for, and it is most of what our approach is built around: a fractional CFO who reads the report, then goes and asks the questions the report could not. The reporting is the instrument panel. Somebody still has to look out of the window.