Finance Transformation IntelligenceFinance Operating Model › Benchmark Insight

Benchmark Insight  Finance Operating Model

What "Good" Looks Like — Reading Finance Benchmarks Without Fooling Yourself

>

IndependentEvidence-backedReviewed Jul 2026Sources 54Methodology

Capability Maturity & BenchmarkingProductivity & Cost of FinanceDecision Making Under Uncertainty

What this publication is for

>

Questions this answers

  1. What are the actual distributions, and against which peer group?
  2. Why is the median the wrong target, and what should I aim at instead?
  3. What does a wide interquartile range tell me that a median cannot?
  4. How do I know whether a benchmark applies to my organisation at all?
  5. How do I turn a gap into a decision rather than a slide?
How to read this.  Layer 1 — Executive summary (4 minutes): the answer and the decision guidance.  Layer 2 — Evidence & analysis: the reasoning, market structure, methodology and references — for controllers, transformation leads and analysts.

Benchmarks are the most misused instrument in finance management. Not because the numbers are wrong, but because of what happens to them in a slide: the median becomes a target, the spread disappears, the peer set goes unstated, and a distribution that described a population becomes an objective for a specific organisation it may not describe at all.

This piece does two things. It publishes the distributions we maintain, with their peer set and method attached. And it sets out the reading discipline that turns them into decisions.

The distributions

Peer set: 58 organisations — mid-market B2B software companies of comparable revenue and stage. Classified independent, high confidence, as of 1 January 2026.

BenchmarkMedianTop quartileInterquartile rangeDirection
Days to close the books645–8Lower is better
Days sales outstanding473840–55Lower is better
Finance cost, % of revenue1.1%0.8%0.9–1.4%Lower is better
Revenue per finance FTE$9.8M$12.5M$8–12.5MHigher is better
Close automation — routine steps automated55%75%40–70%Higher is better
Planning maturity — FP&A operating modelDriver-basedContinuousConsolidated → Driver-basedHigher is better

The interquartile range is published deliberately. Most benchmark presentations omit it, and the omission is where the misuse begins.

Rule 1 — The median is not a target

The median describes what is normal in a population. Normal is not an objective; it is a description of the middle of a distribution that includes organisations with worse constraints than yours, worse systems than yours, and no active improvement programme.

Three consequences:

  • Aiming at the median aims at being unremarkable, and by the time a programme lands —

typically 12 to 24 months — the median has moved.

  • The top quartile is the honest target, because it describes what is achievable by

organisations broadly like you rather than what is typical.

  • Being at the median is not a finding. It generates no decision. The useful questions

are how far the top quartile is and what separates it — and the median answers neither.

The instinct to target the median comes from a reasonable place: it feels achievable and defensible. But a target that would leave the organisation exactly where the middle of the market already is does not justify a transformation budget, and a board will notice.

Rule 2 — The spread is the finding

A narrow interquartile range means the metric is well understood and largely under management control: most organisations cluster, and distance from the cluster is a performance signal.

A wide range means something quite different, and it is the more common case in finance. It means context dominates outcome. Days to close spans 5–8 at the interquartile — a 60% difference between the first and third quartile — which tells you that entity count, ERP quality, audit intensity and whether statutory and management close are separated matter more than effort or competence.

The practical implication is a reordering. Where the spread is wide, the first question is not "how do we get to 4 days?" but "which structural characteristics put an organisation at 5 rather than 8, and which of those do we have?" A programme that skips that question sets a target it may be structurally unable to reach, and burns credibility reaching for it.

Rule 3 — Interrogate the peer set before the number

A comparison against the wrong population is worse than no comparison, because it produces false confidence in a specific direction.

Ours is stated: 58 mid-market B2B software companies. That is a genuine constraint, and it means these distributions transfer well to organisations with similar transaction profiles and poorly to, for example, a multi-entity manufacturer with heavy intercompany volume or a regulated financial institution with statutory reporting obligations that dominate the close.

Three questions to ask of any benchmark before using it:

  1. Who is in the set, and how many? A benchmark without a stated population is a

marketing number.

  1. When was it taken? Finance benchmarks move slowly, but automation coverage does not.
  2. Who assembled it, and what do they sell? A distribution published by a vendor whose

product improves that metric is not independent evidence, whatever its methodology says.

That third question applies to us as much as to anyone, which is why we publish the peer set, the method and the as-of date alongside every figure, and take no placement fees on any product that would move them.

Rule 4 — Convert the gap into a decision, not a target

A gap is not a plan. The conversion has four steps, and most benchmark exercises stop after the first.

  1. Size the gap in the unit that matters. Not "we are two days above median" but "two

days of close effort across the team, at this cost, delaying reporting by this much."

  1. Identify which structural characteristics explain part of it. Entity count and ERP

quality are not excuses; they are scope. The portion of the gap that is structural requires a different intervention from the portion that is process.

  1. Locate the addressable portion on the maturity spine. A gap explained by an L1

process is closed by standardization, not by tooling — a distinction the benchmark itself cannot make.

  1. Set the target at the top quartile, adjusted for the structural portion, and name

the stage that closes it.

Only after step 4 does a benchmark comparison become a management instrument.

Why we publish the qualitative benchmark too

Quantitative distributions answer how much. They cannot answer why, and a finance function that is 8 days on close needs to know which of several quite different conditions produced that number.

That is what the maturity spine is for: a capability at L1 (Manual) and one at L3 (Automated) can both produce a mediocre close-day number, for entirely different reasons and with entirely different remedies. Reading the quantitative benchmark against the maturity position is what turns a comparison into a diagnosis.

If you remember only three things
  1. Target the top quartile, never the median. The median describes what is typical,

including organisations with worse constraints and no improvement programme. It will also have moved by the time you arrive.

  1. A wide spread is a finding, not noise. Where the interquartile range is wide, context

dominates — so establish which structural characteristics apply to you before setting a number.

  1. Always read a benchmark against the peer set and the maturity position. The number

tells you how far; only the maturity position tells you what closes it.

Exhibit 1 — The six distributions, with spread shown rather than hidden

Where this leaves you. Take the one benchmark where you are furthest from the top quartile, and split the gap into its structural and addressable portions before setting any target. The split is usually the whole insight.


Layer 2

Evidence & connections

The reasoning behind the summary above — market structure, methodology, trade-offs and references, for finance transformation leaders, controllers and analysts.

What this rests on

Methodology →
  • dilynx benchmark models — six curated distributions over a peer set of 58 mid-market B2B software organisations, classified independent, high confidence, as of 2026-01-01
  • The dilynx capability maturity spine, used as the qualitative benchmark alongside the quantitative ones
  • Reasoning and judgement on benchmark interpretation, labelled as such
Where a statement is judgement rather than a measured finding, it is labelled as such in the text. Independent — no paid placements. Rankings are never influenced by commercial relationships. Our independence →

If — and only if — technology is part of the answer here, this is the independently assessed market.

Close ManagementFP&AClose Management Buyer's GuideFP&A Software Buyer's GuideBest Close Management Software 2026Best FP&A Software 2026

Related benchmarks

Benchmark Intelligence →

How performance in this area is measured, and what comparable finance organisations achieve.

Close & ReportingCash & Working CapitalCost of FinanceFinance ProductivityAutomation & AI AdoptionPlanning & Forecasting

Related implementation

Transformation Marketplace →

Where the answer is a partner rather than a product — the specialisms that deliver work in this area.

Finance Operating ModelFinance Modernization

Related assessment

How it works →

The Executive Finance Assessment reads your organisation against the same maturity spine, decision archetypes and benchmark models used across this pillar — so what you read here and what it tells you about What "Good" Looks Like — Reading Finance Benchmarks Without Fooling Yourself are expressed in one vocabulary, not two.

Executive Finance Assessment

What does this mean for your organisation?

This research frames the question in general terms. The Executive Finance Assessment answers it for your finance function specifically — your position, your highest-impact move, and the evidence behind it.

Begins with a free Executive Brief — about five minutes, anonymous, no account. Full assessment €59, one-time. It complements the research; it does not replace it.