What is a good Share of Model benchmark?

There is no universal benchmark, and answers differ by category size and by how you sample. In a category with five major competitors, an even split would be about 20% each, so a share well above that suggests strong AI visibility. In a category with many competitors, a smaller share can still be a strong position given the typical 3-5 brand limit in AI responses. The most important signal is trend direction across repeated runs of the same prompts.

Why does my Share of Model vary so much between AI engines?

Each AI engine has different training data, retrieval sources, and recommendation behaviour. ChatGPT may draw heavily from discussions where your brand is mentioned frequently, while Perplexity's real-time search might surface a competitor's recently published comparison article. Gemini and AI Overviews weight Google's own index and structured data differently than independent AI engines. These differences create natural variation by engine, so record each engine separately.

How quickly can I improve my Share of Model?

It depends on the actions taken and the engines targeted. For engines with real-time retrieval (Perplexity, AI Overviews, AI Mode), content improvements can show up within a few weeks. For engines with slower content ingestion, improvements typically take longer. The fastest gains often come from fixing technical blockers (unblocking AI crawlers, publishing llms.txt) and improving high-authority pages that should already be cited but are hard for AI engines to extract from.

Free tool, not actively developed; no accounts. AEO Platform no longer sells subscriptions.

Use case

Share of Model Tracking

Measure your brand's share of AI-generated recommendations vs competitors.

Quick answer

Traditional Share of Voice metrics measure search rankings and ad impressions, but AI engines recommend only 3-5 brands per response. If your brand is not in that short list, you are invisible to AI-first buyers. Most companies have no established way to estimate their share of AI recommendations or compare it with competitors across engines and query types.

Expected outcomes

What you can achieve

Baseline

share of mentions per engine and competitor

Segments

view by engine and query type

Monthly

rerun to see the trend

3-5

brands per AI response (the visibility ceiling)

How it works

How to approach share of model tracking

Define your category queries

Write the prompts that define your market: recommendation queries, comparison queries, "best for" queries, and category exploration queries.

Baseline your Share of Model

Run each prompt several times in each engine. Count how often your brand and each competitor appear, and calculate a share for each engine.

Identify weak segments

Break the results down by engine, query type, and competitor. The segments where you are underrepresented are your highest-priority improvement targets.

Execute improvement actions

Strengthen your position in weak segments: improve content structure, build citations, add structured data, and fix technical access issues.

Repeat on a schedule

Rerun the same prompts and number of runs each month to measure the effect of your actions and notice competitive shifts.

Report to stakeholders

Summarise the sheet in a simple table of share by engine and competitor, with the method noted, so leadership can read the result as a directional benchmark.

The solution

How to approach it

You can estimate Share of Model with a sampling method. Define a query bank of category prompts: recommendation prompts, comparison prompts, "best for" prompts, and category exploration prompts. Run every prompt in each engine you care about, several times each, and record which brands appear.

The basic calculation is simple. For one engine, divide the number of answers that mention your brand by the total number of answers collected. Do the same for each competitor. Break the result down by engine and by prompt type to see where you are strong and weak. Sample sizes are small in a manual run, so treat the numbers as directional and compare like with like between rounds.

Repeat the run on a fixed schedule, such as monthly, with the same prompts and the same number of runs per prompt. The trend matters more than any single number. If the manual effort grows too large, AI visibility monitoring tools can run a query bank on a schedule and calculate the same ratios.

When you find weak segments, work the causes you control: content structure, third-party citations, structured data, and crawler access. The free AEO Platform AI-readiness check covers the technical-signal step only. It does not compute Share of Model or read live AI answers.

Example queries

Queries to monitor

“What is the best [category] tool?”

“Recommend a [category] solution for my business”

“Top [category] platforms in 2026”

“Which [category] company should I choose?”

“Compare the leading [category] providers”

FAQ