Share of Model

Share of Model (SoM) measures how your brand shows up in the answers that AI engines give to real buyer questions — how often you are mentioned, how often you are recommended, and how much of the "answer space" you own versus your tracked competitors.

It is the AI-answer analogue of Share of Voice in traditional media monitoring.

Important disclaimer

All measurements are made via official provider APIs, never by scraping a consumer app. The answer a logged-in person sees in ChatGPT or Gemini is shaped by personalisation, memory, and model version. Our numbers are a clean, repeatable, apples-to-apples measurement — not a claim about any one person's screen. Every metric view and export carries this disclaimer:

Measured via official APIs on <date> with <model>. Consumer apps may answer differently (personalisation, memory, different model versions).

The five engines

How a tracking run works

1. Prompt lock — The run locks the prompt-set version, so each run is always tied to a frozen set of questions.

2. Budget check — The estimated cost (prompts × engines × samples) is checked against your plan cap. If it would exceed the cap, the run is refused before any API call.

3. Parallel sampling — Each (prompt, engine, sample) tuple is executed in parallel with per-provider concurrency limits. Results are stored with the full response text, model version, latency, and raw response archived to object storage.

4. Resumable — A unique constraint on (run\_id, prompt\_id, engine, sample\_index) means re-running never duplicates an answer. Partial failures finish as partial runs.

The metrics

All rates are computed over unbranded prompts unless you ask otherwise (a branded prompt that names your brand would trivially mention it). Every rate carries a 95% Wilson score confidence interval.

Reading the charts

• Trend line — Week-over-week mention rate. A rising trend means more AI answers include your brand.
• Engine breakdown — Your share on each engine separately. Some brands perform well on Perplexity but not on Gemini; this tells you where to focus.
• Prompt breakdown — Which tracked prompts you appear in and which you do not. Gaps indicate content to create.
• Competitor comparison — Your share vs. competitors for the same prompts.

What drives Share of Model

• Topical authority — comprehensive, well-cited coverage of your subject area
• Entity clarity — consistent, unambiguous brand signals across the web
• Technical accessibility — AI crawlers can reach and index your pages
• Citation quality — the sources that cite you are themselves trusted sources