A cash flow forecast for a services business starts from people, not from sales. Revenue in any month is the number of billable people, times the days each of them can bill, times the rate for their grade, minus the bench: the people you pay who are not billing. Cash is that revenue delayed by your payment terms and set against a cost base that arrives on the same day every month whether the days were billed or not. Most services founders run a sales forecast dressed as a cash forecast, with utilisation (utilization) assumed rather than planned, and the difference between the two is what carries a company to the edge of a cliff while the pipeline looks excellent.
I am writing from the numbers we ran at DevOpsGroup, a services company helping other businesses build and run software, the services business my co-founder Steve and I started in 2013, as recorded in my notes and diaries at the time. The method below is what I would build now.
Why a services forecast is people times days times rate
A product company can sell more units than it planned. A services company cannot sell more days than it has. That single fact changes the shape of the forecast. Capacity is the ceiling on revenue, and the cost of that capacity is paid in full whether it is sold or not.
So the forecast begins with a headcount plan by grade, month by month, for twelve months. For each person it needs a billable-days assumption: working days in the month, less holiday, sickness, training, selling and the internal work that keeps the firm running. Then a rate per grade, taken from your rate card, and the utilisation you are planning for, not the one you are hoping for.
Multiply through and you have capacity revenue: what the firm would bill if everyone billed at plan. Show the bench as its own line underneath, the paid days nobody billed, at cost. Do not bury it inside a utilisation percentage. A board that sees the bench as a number in pounds each month has a decision to make; a board that sees a utilisation percentage has a slide.
Then convert revenue to cash on the terms clients actually keep, not the terms in the contract, and set it against the cost base by month: salaries, contractors, rent, software, insurance, all arriving on their dates regardless.
A board pack should show three revenue lines for each month, not one. Capacity revenue, if everyone billed at plan. Forecast revenue, at the utilisation you have planned. Contracted revenue, from signed work only. The gap between the second and the third is what sales has to close, and the question of when it can be closed is the subject of the rest of this page.
The sustainable ceiling is not your best month
By late 2019 our monthly revenue had risen a long way. It had also stopped rising reliably enough to carry the shape of the company around it. We could reach a level of revenue occasionally, and the cost base had been built for a company that reached it every month. When somebody asked “what if we sell more?”, the answer had to include how often we could do it. One excellent month does not change what the company costs in the next one.
The best month, a high point after investment, was evidence of capability. It was not evidence of a run rate. A cost base does not ask for a good month. It asks for another one, and then another.
The test I would run now takes twelve months of revenue, strips out the one-off work that will not repeat, and finds the level the firm reached in at least half of those months. That is the ceiling. Set it against the direct-cost line and the overhead line. If breakeven sits above the ceiling, the firm does not have a sales problem that another push will fix. It has a shape problem: too many kinds of work, too much cost, or a rate card that does not carry the bench. Every additional sales target still depends on somebody outside the business making a decision on their timetable, and effort on your side does not give you control of the date. Simplifying what you sell, and what it costs to deliver, is the lever you do control. It felt like a retreat at the time. It was the way to build from a month we could repeat.
Pipeline coverage is two questions, not one percentage
The pipeline coverage ratio is the weighted pipeline for a period divided by the revenue you need in that period. In December 2019 ours covered the following January and February comfortably. On paper there was enough potential work, adjusted for the likelihood of winning it, to support the revenue we expected. Early 2020 showed the problem with the number. An opportunity can be real, a customer can be interested, and the work can still begin later than the month in your forecast. The pipeline contained two different questions, and the coverage percentage made them look like one.
Will we win it? When can it actually produce revenue?
For the first, you want evidence that the customer intends to buy from you: a named budget holder, a procurement route, a competitor eliminated. For the second, you want evidence about the remaining steps, who controls them and how long similar steps have taken before. A probability attached to a sales stage is not evidence of a start date.
Take an invented example. A £100,000 project with a seventy per cent chance of being won goes into a weighted pipeline as £70,000. If the contract is signed in the last week of March, very little of it can be delivered in March, and the first invoice will be paid on terms after that. The £70,000 cannot be treated as March cash because the opportunity sits under March in a report. The figures are made up; the mechanism is not.
For each material opportunity, record three dates: the earliest credible start, the most likely start, and the latest start the company can afford to assume. Keep the history when those dates move, and ask what evidence changed. A date that advances because the customer completed a step is different from a date that advances because another month has passed in your forecast.
Then report coverage twice. Weighted coverage, by probability of winning, is the sales team’s number. Start coverage, counting only the opportunities with evidence of a start in the month, is the number the cash forecast uses. The firm is safe when the second covers the gap. The first tells you whether the second might improve.
Sales forecast vs cash flow forecast
In early 2020 our coverage percentage answered the sales question while payroll was asking the cash question. The two forecasts answer different questions and belong to different people.
| Sales forecast | Cash flow forecast | |
|---|---|---|
| Question it answers | Will we win it? | When does it start, when is it invoiced, when is it paid? |
| Unit | Bookings, at signature | Cash in the bank, by month |
| Owner | Sales | Finance, with delivery |
| Coverage measure | Weighted pipeline against target | Start coverage against the cost base |
| What it ignores | Start dates, bench, payment terms | Nothing it can afford to |
One coverage percentage asks the sales team, delivery, finance and the board to share an assumption they each understand differently. Separated, the sales team can work on the likelihood of winning, delivery can prepare for a realistic start, finance can model the timing, and management can decide which commitments are safe before the uncertainty clears. That last decision, which hires and which costs to commit to ahead of revenue, is the one the forecast exists to inform.
Building the forecast: the method in six lines
- Headcount by grade, by month, for twelve months, including planned hires with their start dates.
- Billable days per person per month after holiday, sickness, selling, training and internal work, taken from your own history rather than a round number.
- Rate per grade from the rate card; revenue at planned utilisation; contracted revenue shown separately.
- The bench as its own line: paid unbilled days at cost.
- Cash on actual payment terms, using your history of when each client pays.
- The cost base by month, the twelve-month ceiling, and the gap between them, with start coverage against that gap.
That is board arithmetic. It fits on one page and it is enough to see the gap months before it arrives. The pipeline coverage ratio playbook, in development, sets out the ceiling test and the two-question reading of the pipeline in full, and the cash forecast model, also in development, finds the sustainable ceiling in twelve months of your own numbers and sets it against your own cost lines.