The KPIs an agency or consultancy should carry on a monthly scorecard are nine financial measures: revenue, the split between project and recurring revenue, pipeline coverage, cost of sales, revenue per billable person, cost per billable person, gross margin, overheads and operating profit. Each sits beside the comparable period and the plan, carries a written definition, and answers one question. Utilisation (utilization) is not one of the nine. It is an input to one of them, and reporting it on its own is the most common way a services firm persuades itself that a busy month was a good one.
That list comes from a note I wrote in February 2020. In 2013, my co-founder and I had started the business that became DevOpsGroup, a services company helping other businesses build and run software. Seven years on, I wanted a picture of the company from one page rather than from several reports and a round of questions. The note asked for two levels, company and division, and named the nine measures. What follows is that list, the question I would now attach to each measure, and what I learned from watching a correct dashboard report the same numbers for months without moving any of them. I draw on my notes, diaries and recollections, checking the dates and figures as I write.
The nine measures and the question to ask of each
A one-page dashboard is easy to request. The useful work is deciding what deserves the space. These are the nine that earned it, and the question each one exists to answer.
| Measure | The question it answers |
|---|---|
| Revenue | What work is actually becoming income? |
| Professional services and recurring revenue | What kind of income are we building? |
| Pipeline coverage | What future work supports the plan, and how uncertain is it? |
| Cost of sales | What does delivering the work cost? |
| Revenue per billable person | How effectively is delivery capacity becoming revenue? |
| Cost per billable person | What does that capacity cost us? |
| Gross margin | What remains after direct delivery costs? |
| Overheads | What else must the business support? |
| Operating profit | Does the combined model work? |
Three things belong beside every line. The current period, the comparable previous period and the plan, so that a movement can be read as seasonal, structural or off-plan. A written definition, because if two divisions classify delivery costs differently their margins are comparable only in appearance. And one sentence explaining the largest movement and the decision it creates. If no decision follows, the page should say whether you are watching an outcome, testing an assumption or waiting for something outside your control.
The second line is the one most agencies leave off. Splitting revenue between project work and contracted recurring work every month is how you find out whether the business you are building is the one you said you were building. The recurring revenue entry covers the four ways of counting the second half; the scorecard should say which one it uses.
Why utilisation alone misleads
Revenue per billable person was the one measure I bought rather than invented. We took it from Service Performance Insight’s professional services maturity benchmark, and their published figure became the number our own targets were checked against. The credit for the metric belongs to SPI, not to us.
What we added was a subtraction. We wrote it as: revenue per billable person equals utilisation times pricing, minus what we called dark matter. Dark matter was the name we gave to everything that turns a busy team into an unprofitable one: non-billable work booked as delivery, off-piste pricing agreed in a sales conversation, budget overruns absorbed without a change order, and people moved between teams without the revenue following them.
Utilisation is the first factor in that equation, and only the first. A team can be fully booked at a rate below the card, on a fixed price that overran, with a fortnight of unbilled rework in the middle, and its utilisation report will look excellent. Utilisation up and margin down is a pattern I saw more than once, and a utilisation KPI on its own cannot show it. Read it as an input, put the achieved rate beside it, and treat the gap between utilisation times pricing and the revenue that actually arrived as the measure of how much effort is not becoming money. The utilisation rate entry sets out the formula and where it breaks; the revenue per billable person playbook, in development, works through the four leaks.
The three fixed headers
The nine measures answer the money questions. Later we reported them under a wider frame, and it is the frame I would build first.
Every unit reported on one page under three fixed headers in the same order: Happier Customers, Healthier Teams, Stronger Business. Under each header sat a small fixed set of measures, each with a target, an actual and a trend. Beneath the measures sat a SOFT grid, Successes, Opportunities, Failures and Threats, with a separate block for threat mitigations and an owner against each. SOFT is borrowed, not invented here. The letters are older than SWOT, and the grid is usually credited to the corporate planning research at the Stanford Research Institute in the 1960s that SWOT itself grew out of. We used it because Successes and Failures turn a position audit into a period report, and force failures to be named rather than folded into weaknesses.
The headers earn their place for one reason. They make it impossible to fill a scorecard with money alone. A services firm’s revenue is its people’s time bought by its customers, so a customer measure and a people measure are leading indicators of the nine financial ones, and they belong on the same page in the same month. The financial nine live under Stronger Business. The other two headers each carry two or three measures the unit can own, such as a satisfaction score or renewal, and an employee net promoter score or unplanned attrition. The ancestor of the idea is Robert Kaplan and David Norton’s balanced scorecard; the three headers were our own working version of it, with the borrowed SOFT grid beneath.
The seven selection criteria for what goes under a header are in the monthly scorecard playbook, which is in development. The first criterion is the one that does most of the work: it fits on one page. If it does not fit, it is not the scorecard.
Candidate KPIs and what each one hides
Most lists of agency metrics are long because nothing was refused. This is what the usual candidates conceal when they are reported alone.
| Candidate KPI | What it hides on its own |
|---|---|
| Utilisation | The rate achieved, the overruns absorbed and the unbilled work. Busy is not the same as paid. |
| Revenue growth | Whether the growth is margin or volume, project or recurring, and whether the cost base grew faster. |
| Headcount | Nothing about whether the new people are billable, on the bench, or hired for revenue that has not arrived. |
| Average day rate | Discounts negotiated off the card, and the mix of grades doing the work. A blended rate that rose because seniors are doing junior work is not good news. |
| Win rate | The size and margin of what was won, and how many opportunities were never qualified in. |
| Pipeline value | How much of it is real, when it lands, and whether one deal is most of it. Read pipeline as coverage against the plan, as two questions, not one total. |
| Monthly recurring revenue | Which of the four readings it is: contracted, committed, habitual or merely repeated. Say which. |
| Net promoter score | Who answered, and which clients did not. A good score from a shrinking base is a warning. |
| Revenue per employee | The share of employees who are billable. A firm that adds sales and delivery managers sees this fall while revenue per billable person holds. Use the billable version. |
| Client count | Concentration. Twelve clients with one at half of revenue is a very different firm from twelve clients at a twelfth each. |
None of these is useless. Most of them are inputs to one of the nine, and the right place for them is behind the page, in the working that supports a line, not on it.
What the scorecard should not carry
Cash. Keep a separate cash forecast. The nine measures describe performance; they are not a picture of payment timing or available liquidity, and a firm can have a good month on all nine while the bank balance heads towards a cliff. The pipeline coverage ratio playbook covers the two questions the cash view has to answer, and the forecast belongs beside the scorecard, not on it.
A measure with no decision attached. If the page cannot say what changes when the number moves, the number is context, not a KPI. Keep it in the appendix.
Anything that arrived because someone had an interesting number that month. A dashboard that expands whenever a conversation produces a metric eventually becomes another report you have to search.
More than one page per unit. Two levels, unit and company, the same measures in the same place on each, so that the meeting opens on one picture and any two units can be compared without translation.
Benchmarks, and what to do with them
Two external benchmarks are worth knowing by name. For UK agencies, BenchPress by The Wow Company is the annual survey most owners will recognise, and it covers the profit, growth and pay questions an agency board asks. For professional services more widely, Service Performance Insight’s maturity benchmark is where our revenue per billable person target came from.
Use a benchmark to sharpen the question, not to set the target. A target for revenue per billable person should be built from your own cost base and your own margin requirement, then checked against the benchmark to see whether you are asking the impossible. A benchmark used as a target imports someone else’s mix of grades, rates and overheads into your plan. Both surveys publish their own numbers; I am not quoting them here, because the figures change every year and the method does not.
Running the meeting from the page
The page will not make a customer sign or a service become easier to deliver. I learned that quickly. The dashboard I specified in February 2020 reported the company’s revenue ceiling, its breakeven and its margin trough correctly, every month, for months. It moved none of them. A measure you cannot act on this quarter is still worth having, but you should be honest with yourself about which of your nine that describes, and the honest answer for a services firm is usually that cost, margin and revenue per billable person are within reach this quarter, and revenue is not.
What the page can do is make sure the next management meeting begins with the same picture of the company, so that more of the meeting goes on the difficult, useful question: what are we going to change? That is the whole case for a scorecard: the same nine, defined the same way, in the same place, with a decision beside the one that moved most.
Where to go next
The full frame, the seven selection criteria and the meeting structure are in The monthly scorecard, a playbook in development, and the page itself is being built out as a monthly scorecard template for Notion, also in development. The metric behind the fifth line has its own glossary entry. Everything on this subject is gathered under metrics and profitability.