AI Perception Score: how to measure AI's understanding of you
Mention rate tells you whether the model knows you exist. AI Perception Score tells you whether the model actually understands you, and whether buyers reading the answer would come away with the right picture.
- 01AI Perception Score is a composite of recognition, positioning, sentiment, competitor proximity, and source quality, each weighted by category.
- 02It is the difference between being mentioned (a coin flip on whether the answer helps you) and being understood (the answer consistently positions you the way you want).
- 03AI Perception Score is read per model and then aggregated, so a brand can have a strong overall score while still having one or two models that misrepresent it.
- 04By 2027, we expect AI Perception Score and its equivalents to sit alongside organic and paid as a tracked, board-level KPI.
What AI Perception Score measures
AI Perception Score is a single composite score, on a 0–100 scale, that measures how clearly an AI model recognizes, positions, and describes a brand. It combines five inputs: recognition (does the model surface the brand at all on category prompts), positioning (does it describe the brand the way the brand describes itself), sentiment (is the description net positive, neutral, or negative), competitor proximity (which competitors are named in the same answer), and source quality (which domains the answer is cited from).
The point of combining these into one number is to make perception comparable. Mention rate alone is misleading, a brand can be mentioned often but described in a way that costs it deals. AI Perception Score catches that gap.
Why mention rate is not enough
Two brands can have identical mention rates and very different AI Perception Score scores. One is mentioned as the category leader with a clean one-line positioning that matches the brand's site. The other is mentioned as an alternative, with a positioning that subtly mischaracterizes what the product does, alongside the wrong set of competitors.
The first brand is winning the AI answer surface. The second is being damaged by it. Mention rate flattens both into the same number. AI Perception Score is the metric that pulls them apart.
Being mentioned is the easy half. Being understood is the half that decides whether the answer helps you or hurts you.
Get your AI Perception Score score
See how every AI model recognizes, positions, and describes your brand, with the per-model breakdown and the source mix behind each score.
How it is calculated
Each input is scored from the model's answer to a fixed prompt set. Recognition is straightforward: does the brand appear and in what position. Positioning is scored against the brand's own canonical one-line description. Sentiment uses a calibrated sentiment classifier tuned for B2B language. Competitor proximity counts which competitors are named alongside and at what position. Source quality scores the domains cited.
The five inputs are then weighted by category. In a documentation-heavy category like developer tools, source quality and positioning carry more weight. In a brand-led category like consumer fintech, sentiment and competitor proximity carry more. The result is a single 0–100 score that is comparable across weeks and across models.
How to use the score
Use AI Perception Score as a leading indicator, not a vanity number. Watch the per-model breakdown more than the aggregate, a 70 average that is 90 in ChatGPT and 50 in Claude is a very different operational problem than a flat 70 across the board.
When the score drops on a single model, the cause is usually local: source rotation, a competitor publishing a new comparison, an outdated description. When it drops across multiple models, the cause is upstream: the canonical positioning on your own site has gone stale, or a key reference domain is describing you wrong. The fixes for the two cases look very different, and the metric tells you which one you are in.
8 tracked AI models
Strajist measures perception across these 8 AI models, on the same prompt set, on the same weekly cadence.
Frequently asked questions
Is AI Perception Score a Strajist-only metric?
The name is ours, but the underlying idea, that perception in AI answers needs a composite score, not just a mention count, is becoming standard across the category. We expect convergent definitions over the next 12–18 months, much like share of voice converged in SEO.
What is a good AI Perception Score score?
Above 80 is strong category leadership across most models. 60–80 is solid but with one or two models where you are under-described. Below 60 typically signals that buyers using one of those models will come away with a wrong or incomplete picture of your brand.
How often does AI Perception Score update?
Weekly for most categories, which matches the cadence at which model behavior, source freshness, and competitor publishing patterns meaningfully shift. Daily updates are mostly noise unless you are actively shipping changes and watching the response.
Get your AI Perception Score score
See how every AI model recognizes, positions, and describes your brand, with the per-model breakdown and the source mix behind each score.