Benchmarking AI Maturity
We don't publish industry benchmark figures. A peer average is only as good as the sample behind it, and we don't have one we can document well enough to publish. Here's what a credible benchmark needs, and how to benchmark against yourself in the meantime.
What a Credible Benchmark Tells You
Before you compare yourself with anyone's “industry average” — ours included, if we ever publish one — check that it states these four things next to the numbers.
Who was measured
How many organizations, how they were recruited, and which sizes, sectors and regions they cover. A self-selected sample of enthusiasts says little about an industry.
When and how
The collection period, the questions asked and the scoring method. A score from a different questionnaire is a different measurement, not a comparison.
Self-reported or observed
Whether the figures are answers to a survey or measured from systems. Both are useful; they are not interchangeable.
What the categories cover
Whether the categories add up to the whole sample, and what was left out. A distribution that does not reach 100% should say why.
Benchmark Against Yourself First
Record a baseline
Run the AI Execution Maturity Assessment once, with the people who actually do the work in the room. Export the result.
Re-run on a schedule
Repeat it each quarter with the same people and the same honesty. A rising domain score is evidence; a single score is a snapshot.
Compare inside the business
Run it per team or business unit. Differences between your own groups are comparable because the method, period and definitions match.
The Five Maturity Levels
The AI Execution Maturity Assessment turns your answers into a 0–100 score and maps it to one of five levels. These bands are PromptFluent's own scale, not an external standard.
| Level | Score | What it looks like |
|---|---|---|
| 1 · Experimental | 0–19 | Ad hoc AI usage with no governance or measurement |
| 2 · Fragmented | 20–39 | Scattered policies with inconsistent implementation |
| 3 · Structured | 40–59 | Defined controls for key use cases with baseline measurement |
| 4 · Governed | 60–79 | ISO 42001-style discipline across AI lifecycle |
| 5 · Compounding | 80–100 | Continuous improvement with measurable competitive advantage |
Scoring details are on the methodology page.
Get Your Baseline
Run the AI Maturity Assessment to get your level and domain scores — the first point on the only benchmark you can fully trust: your own trend.