Finance cost as a percentage of revenue is the most widely used justification for finance transformation, and one of the least useful measures of whether it worked. Our own benchmark puts the peer median at 1.1% and the top quartile at 0.8%, and those numbers are real. The argument here is not that the measure is inaccurate. It is that it is the wrong thing to optimise, and that organisations which optimise it directly frequently end up worse.
To be clear about what is being claimed: this is a position, argued from how the measure behaves, not a finding derived from data. It should be read as such.
The structural problem
Cost of finance can be improved by doing less. Any measure that improves when you reduce output is a constraint, not an objective — and treating it as an objective inverts the thing you were trying to manage.
Every route to a better cost ratio is available without improving anything:
- Stop doing work. Reduce reporting frequency, cut analysis, shrink the reconciliation
population. The ratio improves immediately. The consequence arrives in a later period, in a different function, attributed to something else.
- Defer investment. System replacement, control remediation and data work all raise cost
in-year and lower it later. Deferring them improves this year's ratio and mortgages the next five.
- Move cost elsewhere. Push reconciliation into the business, invoice queries into
procurement, reporting into a BI team. The finance line falls; organisational cost does not.
- Grow revenue. The denominator does the work. This is the version that flatters
management most and reflects finance performance least.
Three of those four make the organisation worse. All four improve the metric.
What the ratio cannot see
The deeper problem is that the measure treats a judgement function as a processing function. Cost per unit is a coherent measure where units are comparable — transaction processing genuinely qualifies. But a large and growing share of what finance is for has no unit: whether a forecast was believed, whether a pricing decision was informed, whether a risk was surfaced before it became an event.
A finance function that costs 0.8% of revenue and produces forecasts nobody trusts is performing worse than one at 1.3% whose numbers are relied on. The ratio ranks them the other way round, confidently.
| Measure | What it rewards | How it is gamed | Sees capability? |
|---|---|---|---|
| Finance cost % of revenue | Spending less | Do less · defer investment · push cost to other functions · grow revenue | No |
| Revenue per finance FTE | Carrying more per person | Outsource (cost moves, headcount falls) · defer hiring into a gap | Partly |
| Effort split: run vs. change | Capacity to improve | Reclassify run work as change | Yes |
| Reliability: forecast accuracy, restatements, audit adjustments | Being right | Hard to game — the outturn arbitrates | Yes |
The right-hand column is the argument. Only the bottom two measures move when the finance function actually gets better at its job.
What to measure instead
Two measures, held together. Neither is sufficient alone, and the pairing is the point.
1. Capacity — the run/change split
What share of finance effort is spent keeping the lights on versus improving how the function works?
This is the measure that predicts the future. A function at 95% run has no capacity to absorb an acquisition, a system migration or a regulatory change, and every such event will be absorbed by overtime and error. A function at 75% run can improve while operating.
Cost reduction that improves the ratio by removing change capacity is visible here immediately, and invisible in the cost line. That is exactly the asymmetry that makes it worth measuring.
The objection — that the split is estimated rather than measured — is fair and does not defeat it. A quarterly estimate by process owners, tracked consistently, is directionally reliable enough to manage against, and the act of estimating it changes behaviour.
2. Reliability — is the output trusted?
Forecast accuracy against outturn. Restatements. Audit adjustments. Late reporting.
These are the measures a finance function cannot game, because reality arbitrates. They are also what the rest of the organisation actually experiences of finance: not what it costs, but whether its numbers can be relied on when a decision depends on them.
Reliability is where cost reduction shows up as damage, on a lag of two to four quarters — which is precisely why it must be measured alongside cost rather than after it.
Then keep cost as a constraint
None of this argues that cost does not matter. It argues for a different role in the sentence.
Objective: increase capacity and reliability. Constraint: without increasing cost as a percentage of revenue.
That framing is harder to satisfy and much harder to fake. It is also the framing under which technology investment gets evaluated properly: automation that reduces headcount while reducing reliability fails the objective and passes the constraint — which the cost ratio alone would score as a success.
Our own benchmarks support holding both: the top quartile sits at 0.8% cost and $12.5M revenue per finance FTE. Those organisations are not cheap; they carry more per person. The distinction between being cheap and being productive is exactly what a single ratio cannot express.
The strongest case against this argument is that cost is the only measure with genuine external discipline, and the alternatives are soft enough to become an excuse.
This deserves to be taken seriously. Capacity split is self-reported and self-serving — every function believes it is at 90% run and needs more people. Reliability measures are lagging, sometimes by quarters, and are easy to attribute elsewhere ("the forecast was wrong because sales changed the pipeline"). Cost, by contrast, is unambiguous, externally comparable, and cannot be argued with. A CFO who replaces a hard metric with two soft ones has, from the board's perspective, replaced accountability with narrative — and boards have seen that move before.
There is also a legitimate case for cost as a primary objective in specific situations: a function that is genuinely bloated relative to peers, a post-merger integration with duplicated structures, or a business under real liquidity pressure. In those cases the cost ratio is measuring the thing that actually needs to change, and elaborate arguments about capacity are a way of not doing it.
What survives the objection is narrower than the headline: cost is the right objective when the function is demonstrably out of line with peers, and the wrong objective once it is within range. At 1.4% against a 1.1% median, cost is a legitimate target. At 0.9%, continuing to optimise it is optimising the wrong variable, and the honest version of this argument says so rather than defending every budget in every circumstance.
Making the case without sounding defensive
The credibility problem is real: a CFO arguing that cost is the wrong measure sounds like a CFO defending a budget. Three things make the argument land.
- Accept the constraint explicitly and first. "We will not increase cost as a share of
revenue" removes the suspicion before the argument starts.
- Bring the measures, not the objection. Arriving with a capacity split and a
reliability trend is a different conversation from arriving with a reason the cost target is unfair.
- **Name where cost is the right target.** Conceding the cases in the counter-argument
above is what makes the rest credible.
- Any measure that improves when you do less is a constraint, not an objective. Three
of the four routes to a better cost ratio make the organisation worse.
- Measure capacity and reliability. The run/change split predicts whether the function
can absorb what is coming; reliability is the only measure reality arbitrates.
- **Cost is the right objective when you are out of line with peers, and the wrong one
once you are within range** — the honest version of this argument concedes that.
Where this leaves you. Before the next budget cycle, produce one estimate of your run/change split and one reliability trend. If you have neither, the cost ratio is currently the only thing anyone can see — which is why it is the only thing anyone manages.