Eight numbers tell you almost everything about an agency's health: gross margin, net margin, utilization, realisation, project margin spread, debtor days, work in progress, and cash runway. Review them monthly, in that order, and look at the trend rather than the absolute figure. The one most agencies are missing is project margin spread, because a healthy company-level margin routinely conceals two or three accounts that are losing money - and the aggregate tells you there is a problem while the spread tells you where.
Most agency owners review revenue monthly, profit quarterly, and everything else never.
That is not carelessness. Revenue is easy to see and the rest requires assembling data from three systems that do not agree. So the review becomes "did we bill more than last month", which is the one number that tells you least about whether the business is healthy.
An agency can grow revenue while margin falls, cash tightens and two accounts quietly lose money. All three are visible in numbers you already have.
This guide covers the eight metrics worth reviewing monthly, what each one tells you, what to do when it moves, and how to build a review that takes twenty minutes.
Why revenue is the least useful number
Revenue is the figure every agency owner knows and the one that tells you least. An agency can grow revenue while margin falls, cash tightens and two accounts quietly lose money - and the revenue line will look excellent throughout.
The eight below are the numbers that would have shown you each of those, and every one of them can be calculated from data you already collect.
Read them in this order
Gross margin first, because it tells you whether the core activity works. Cash runway last, because it tells you how urgent everything above it is.
The eight
In the order worth reading them.
1. Gross margin
Revenue minus the direct cost of delivery, divided by revenue. Direct cost means the people doing billable work - not rent, not software, not operations.
The most important number in an agency, because it tells you whether the core activity works before any overhead. Target 50% or better. An agency at 30% cannot overhead its way to profitability; there is not enough left.
When it falls: the cause is almost always underpricing, absorbed scope, or seniority drift - work priced for mid-weight people delivered by senior ones. Covered in agency profit margins.
2. Net margin
What is left after everything. Target 15-25%.
Read it alongside gross margin, because the two together locate the problem. Healthy gross and poor net means your overhead is too heavy for your size. Poor gross means the problem is upstream in pricing or delivery, and cutting overhead will not fix it.
The correction to apply before comparing to any benchmark: is owner time costed at market rate? If you are doing billable work and paying yourself below replacement cost, the margin is flattered by exactly that gap.
3. Utilization
Billable hours ÷ available hours. Available meaning genuinely available - after holiday, internal work and admin, which for most roles is 25 to 32 hours a week rather than 40.
Read it as a diagnostic rather than a target. Consistently low means you have sold too little or your available-hours figure is wrong. Consistently very high means no slack to absorb anything, which is a delivery risk and a burnout risk.
Look at the distribution, not the mean. A team averaging 70% where everyone sits between 65 and 75 is healthy. The same average with two people at 95 and two at 45 is not, and the average conceals precisely the problem you need to fix. Our guide to utilization rate covers the competing formulas and why the denominator matters.
4. Realisation
Invoiced hours ÷ billable hours tracked.
The number most agencies have never calculated and the one with the fastest available payback. Anything under 100% means work was delivered, logged as billable, and then written off - through discounting at invoicing, unbilled overage, or scope absorbed without a change order.
Target 95%+. At 85%, one hour in seven of your delivered work was free, which is equivalent to a 15% price cut you never decided to make.
High utilization with low realisation is the most dangerous combination in a services business: everyone is at capacity, everyone is exhausted, and the revenue does not reflect it.
5. Project margin spread
Gross margin per project, viewed as a distribution.
The metric most likely to be missing and most likely to change a decision. Company-level margin is an average, and averages hide the thing you need to see.
A typical agency at a respectable 22% net looks like this underneath: a few projects at 60% gross, most around 45%, and two or three at or below zero. The good work subsidises the bad indefinitely, because nobody calculates it per project.
Target: no project below 30% gross. More usefully, look for the pattern - which client type, service line or project size keeps appearing at the bottom. That pattern is the actionable output.
6. Debtor days
Average days from invoice to payment.
If your terms are 30 and this is 52, you have a collection problem worth more than most cost savings, and it is fixable with process rather than negotiation.
When it rises: usually a chasing problem rather than a client problem. Most late payment is administrative - an invoice sitting in an approval queue. Our guide to cash flow covers the escalation sequence.
7. Work in progress
Work delivered but not yet invoiced.
One of the earliest available warnings, typically visible a month or two before a cash problem arrives. A rising WIP balance means you are delivering faster than you are billing.
Watch the trend and the age. A stable balance is fine; 15 days rising to 28 over a quarter is a signal. Anything over 60 days old should be treated as at risk rather than as an asset - see work in progress.
8. Cash runway
Months of operating cost in the bank, excluding money owed to tax authorities.
Three months is a reasonable floor for a services business. Below one is an emergency regardless of how strong the pipeline looks, because pipeline does not pay salaries.
The pairing rule
No metric here is trustworthy alone. Utilization without realisation flatters. Margin without the project spread conceals. Cash without WIP misses what is coming.
Read them in pairs and the picture is reliable. Read any one in isolation and it will eventually mislead you, usually in the optimistic direction, because the errors that produce these numbers all come from work that was done and not recorded.
Two more worth watching quarterly
Client concentration. What share of revenue comes from your largest client. Above 30% and their payment behaviour is your cash flow; above 50% and their procurement department effectively runs your finance function.
Revenue per head. Total revenue ÷ total headcount, including non-billable. A blunt but useful measure of whether the structure is proportionate to the business. Falling revenue per head during growth usually means overhead arrived before the scale to carry it - common and temporary between roughly six and fifteen people.
Trend beats absolute
Every number here is more useful as a direction than as a value. A 44% gross margin means little in isolation; 44% having been 51% two quarters ago is a finding with a cause worth locating.
This is also why benchmarking against other agencies is mostly a distraction. Comparability is poor - the biggest variable in any reported margin is how the owners pay themselves - and your own trajectory is both more accurate and more actionable.
Reading them together
Individually these are numbers. In combination they diagnose.
Revenue up, margin down. You are buying growth with price. Check realisation and project margin spread; you are probably discounting to win.
Margin fine, cash tight. A billing problem, not a profitability one. Look at WIP and debtor days - you are delivering and not invoicing, or invoicing and not collecting.
Utilization high, margin low. Everyone is busy on work that is underpriced or over-serviced. Check realisation first, then project margin spread.
Utilization low, margin fine. A sales problem rather than a delivery one. The work you do is profitable; there is not enough of it.
Everything fine, people exhausted. Look at the utilization distribution rather than the average. The load is unevenly spread, and the aggregate is hiding it.
That last one is worth emphasising, because it is the case where the finance review says everything is healthy and the team knows otherwise.
The prerequisite nobody mentions
Every metric here except cash derives from tracked time. If time is entered late, incompletely, or reconstructed from memory on Friday, the whole review is decorative - and the errors all run in the same optimistic direction, because unrecorded work is always work you did and were not paid for.
Getting time entered within two days is not a reporting improvement. It is the thing that makes reporting possible at all.
Building the review
Twenty minutes a month if the data is clean. The data being clean is the actual work.
Three prerequisites:
Time entered within two days. Every metric except cash derives from tracked time. Timesheets reconstructed on Friday make the whole exercise decorative - and reconstruction is always low, so the errors run in one direction.
A real fully-loaded hourly cost per person. Salary plus employment costs divided by genuine available hours, as covered in calculating a billable rate.
Invoices raised on a schedule with a named owner. Otherwise WIP and debtor days measure your admin habits rather than your business.
The review itself:
Same day each month. One page. Each metric with its current value, last month's, and a direction arrow. Then one question: which number moved most, and why?
Not a report to circulate. A twenty-minute look, by whoever can act on it, at a page that takes ten minutes to assemble.
Starting from nothing
If you currently track only revenue, the sequence that gets you to a useful review fastest.
Month one: gross margin. Total revenue against the fully-loaded cost of the people who delivered it. Even approximately, this is the number that tells you whether the core activity works.
Month two: realisation. Compare tracked billable hours to invoiced hours for a single month. This usually produces the biggest surprise and the fastest available improvement, because closing the gap requires no client conversation at all.
Month three: project margin on your last ten projects. One hour. Almost always reveals a pattern - a client type, service line or project size that consistently loses money.
Month four: cash runway and debtor days. Both are straightforward once you look, and together they tell you whether a margin problem is also a survival problem.
Then add the rest as the underlying data improves. Utilization and WIP both depend on time being entered promptly, which is a habit rather than a calculation and takes longer to establish.
Four months to a real financial picture, starting with the two numbers that most often change a decision. Attempting all eight in month one usually produces a spreadsheet nobody maintains.
Reviewing them as a team, not alone
Most agency owners look at these numbers privately, which halves their value.
Share gross margin and utilization with the delivery team. Not as a performance measure - as context. A team that knows a project is fixed-fee at a particular margin makes materially better decisions about absorbing an extra request, because they understand what it costs. A team told nothing has no reason to treat scope as finite, and then gets blamed for over-servicing.
Share the project margin spread, anonymised if you prefer. The pattern of which projects lose money is usually visible to the delivery team before it is visible in the accounts, and asking them why a particular project type keeps landing badly produces better diagnoses than any spreadsheet.
Keep cash runway with whoever can act on it. This one is genuinely leadership-only, because it produces anxiety without agency in anyone who cannot change it.
The general principle: share the numbers people can influence, and be explicit that they are used for capacity and pricing decisions rather than individual evaluation. Say it more than once, because people will assume otherwise until they have seen it be true.
Turning a number into a decision
A metric that does not change behaviour is a number you are collecting for its own sake. Each of the eight has a specific decision attached, and it is worth being explicit about which.
Gross margin decides whether your pricing works. Falling means raise rates, tighten scope, or check seniority drift.
Net margin decides whether your overhead is proportionate. Healthy gross with poor net means the structure, not the pricing.
Utilization decides hiring. Sustained overrun with pipeline means capacity; without pipeline it means estimation.
Realisation decides whether to fix billing discipline before anything else. Under 90% it is almost always the fastest available improvement.
Project margin spread decides which work to stop taking, re-price or re-scope.
Debtor days decides whether to change your chasing process or your terms.
WIP decides whether to move to milestone billing.
Cash runway decides whether any of the above is urgent.
If a number moves and no decision follows, either the movement was noise or the metric is not earning its place. Both are worth noticing.
What each metric looks like when it is lying to you
Every one of these can be technically correct and materially misleading. Knowing how is most of reading them well.
Gross margin looks healthy because delivery cost is understated. Almost always because owner time is uncosted, or because time was entered late and incompletely. Both flatter the number in the same direction.
Utilization looks healthy because the denominator is wrong. Available hours calculated at 2,080 rather than the real figure produces a comfortable-looking number that means nothing.
Realisation looks fine because out-of-scope work was never logged. If a request is absorbed without ever being recorded as billable time, it does not appear in the realisation calculation at all - the work was free and the metric shows 100%. This is the most insidious of the four, and the reason change orders matter to measurement as well as to margin.
Cash runway looks fine because tax money is sitting in the main account. VAT and payroll tax collected but not yet paid is not working capital, and treating it as such is one of the more common ways an otherwise healthy agency runs into trouble.
The common thread is that every distortion runs in the optimistic direction. That is not coincidence - it is that the errors come from work not being recorded, and unrecorded work is always work you did and were not paid for.
Where the data comes from
The most common obstacle is not knowing which metric to track. It is that the numbers live in three systems that disagree.
Four practical fixes, in order of impact.
One source of truth for time. If hours live in a tracker, project margin lives in a spreadsheet and invoices live in accounting software, reconciliation is a monthly manual job that will be abandoned by March. The closer these are to one dataset, the more likely the review actually happens.
Fully-loaded cost recorded per person, not per role. Roles average; people differ, and project margin calculated on role averages is systematically wrong on any project with an unusual staffing mix.
Invoices linked to the work they cover. Otherwise realisation and WIP are guesses. This is the link most commonly missing, and without it two of the eight metrics are unavailable.
A fixed monthly close. A day each month when the numbers are considered final. Without it you are comparing a settled month against a partial one and drawing conclusions from the difference.
None of this requires sophisticated tooling. It requires the data to be entered close to when the work happened and to be connected, which is a process question rather than a software one.
Start with two
If eight feels like a lot, start with gross margin and realisation. The first tells you whether the work is economically viable; the second whether you are actually being paid for what you delivered. Between them they catch most of what goes wrong.
What not to do
Do not track all eight from day one if you have none. Start with gross margin and realisation. Both are calculable this month and both usually reveal something actionable immediately. Add the others as the data improves.
Do not benchmark obsessively. Published figures are orientation, not targets, and comparability is poor - the biggest variable in any agency's reported margin is how the owners pay themselves. Your own trend is more informative than anyone else's number.
Do not turn utilization into a performance metric. The moment people are judged on it, timesheets start reflecting expectations rather than reality, and you lose the ability to trust every number derived from them.
Do not review without deciding anything. A metric review that changes no decisions is a ritual. If nothing would alter based on a number, stop tracking it.
One page, once a month
That is the whole practice: eight numbers, one page, twenty minutes, same day each month, looked at by whoever can act on them.
The summary
Eight numbers, monthly, on one page: gross margin, net margin, utilization, realisation, project margin spread, debtor days, WIP, cash runway.
If you track two, make them gross margin and realisation - the first tells you whether the work is economically viable, the second whether you are actually being paid for the work you did. Between them they catch most of what goes wrong in an agency.
And look at the project margin spread at least quarterly, because the aggregate will tell you there is a problem long before it tells you which client it is.