A forecast that held for a full quarter, on flat revenue
A marketing agency with two service models, direct and white labelRevenue roughly flat year over year, profitability up, and a forecast that came in nearly identical to actuals for a full quarter.
Percentages, multiples, counts and timeframes only. Client revenue, profit and cash figures stay private.
The situation
The agency runs two service models: direct client work and white label work for other agencies. Two owners, both deep in delivery.
At the end of 2025 the honest assessment was that a very busy year had left "zero margin for company growth," and that the owners needed to step away from daily operations for the company to become more valuable. That's a structural problem, and no amount of working harder solves it.
By the following spring there was a second tension worth naming, because it's the one nobody puts in a case study. A leaner team made cash flow more predictable. It also made bandwidth thinner and the work heavier. As one owner put it, having almost no employees looks good on paper and increases the workload. Both things were true at once.
What we did
Built the 2026 forecast from the bottom up
Not a growth rate applied to last year. Recurring revenue split by service model, then each model split into small and large accounts by customer count and average price, with large defined at a specific average price point. Project revenue forecast separately from recurring.
That structure is what makes the forecast arguable. When the owners disagreed with an assumption, they could point at the assumption instead of at the total.
Modelled losing the largest account
One large direct account had a new decision maker who was shopping for something cheaper. Rather than note the risk and move on, we modelled the year with that revenue gone. If it happens, the plan already exists. If it doesn't, the downside case cost an afternoon.
Re-scoped project revenue to the months it actually closes
The prior year's project total was inflated by a large pass-through animation expense, which made the base look better than it was. We cut the projection and squeezed project work into a March-to-October window, because high-value project work doesn't close in December or January. Forecasting it evenly across twelve months had been building in a miss every Q1.
Picked the conservative acquisition assumption
On the white label side, two new small customers a quarter. On the direct side, the owners were aiming for a net new customer every month once the new staffing structure was in. We modelled net one per quarter to start, ramping to one a month from Q2, because January and February are reliably slower. A marketing spend line went into the forecast alongside it.
Split cost of goods sold by service model
Direct work and white label work have different economics, and blended COGS hides which one is carrying the business. Splitting it by percentage made each model's real margin visible.
Reforecast when the facts changed
The following spring the forecast was rebuilt around client cancellations and a compensation change, and we benchmarked the cost of a new account manager role across pay bands before anyone posted it. We also pulled a detailed vendor and software expense list for line-by-line review, which is the least glamorous item on this page and one of the most reliably useful.
Where it landed
Q1 2026 revenue came in roughly level with Q1 2025, with slightly more invoicing and meaningfully better profitability. Three months in a row printed consistent operating profit, and three months in a row produced positive cash profit. Across the period the rolling average share of revenue going to people came down about 3 points.
The forecast itself is the quieter result. Plotted against actuals and against the prior year, all three lines came in nearly identical for the quarter. A forecast that holds for a quarter is a forecast you can make decisions against.
The line that stuck with me came while we were sizing a future support hire against the model, and it was not about the model.
we waited too long to start doing this.
Owner, marketing agency
At the end of that same session: "this was wildly helpful. We're meeting in February as a team." And: "this has been enlightening for sure."
His co-owner made the more precise point. They would have got to splitting the two service models eventually, he said, but seeing the numbers "makes it very tangible for us." That is the argument for building a forecast from the bottom up. It rarely tells you something you could never have worked out. It makes the thing you half-suspected concrete enough to act on this quarter instead of next year.
What this maps to
- Service: fractional CFO for agencies
- OS module: Team Pay System, for people cost as a share of revenue
- OS module: Agency Profit Engine, for operating profit by revenue band
Questions
The ones people actually ask on the first callIs flat revenue a good outcome?
It is when profitability improves and the forecast holds. Revenue came in roughly level year over year while operating profit stayed consistent for three consecutive months, which means the same top line produced more money and did so predictably.
What makes a bottom-up forecast better than a growth rate?
You can argue with it. Splitting recurring revenue by service model and then by account size means a disagreement lands on a specific assumption, like how many new accounts a quarter, instead of on a total nobody can interrogate.
Why model losing your biggest client?
Because the cost of modelling it is an afternoon and the cost of being surprised by it is a year. In this case a large account had a new decision maker looking at cheaper options, so the year was modelled with that revenue removed.
Why does project revenue get forecast separately?
Because it does not arrive evenly. Spreading project work across twelve months builds a miss into every slow quarter. Here it was re-scoped into a March-to-October window, which is when that work actually closes.
How much of this is the forecast and how much is the cost cuts?
Both, and they are hard to separate. Bringing people cost down about 3 points as a share of revenue produced the margin. The forecast is what made the owners willing to make those calls, because they could see what each one did to the year before committing to it.
Next step
A 30-minute call, and you will know whether this fitsMonthly bookkeeping, a Fathom dashboard, and fractional CFO advisory for marketing and creative agencies. Logan works every account, and the rates are published rather than quoted.