404 Models / Resources / Share of Model Calculator

AI visibility

Prompt sample

Tracking

Share of model calculator

Share of model calculator for AI search visibility

Share of model calculator for AI search visibility

Share of model measures the percentage of counted category mentions that belong to your brand inside a controlled sample of AI-generated answers. Enter your sampled counts to calculate a transparent, benchmark-free result.

Share of model measures the percentage of counted category mentions that belong to your brand inside a controlled sample of AI-generated answers. Enter your sampled counts to calculate a transparent, benchmark-free result.

Formula

Brand mentions ÷ total category mentions × 100. The denominator includes your brand plus every competitor count entered.

Sampling rule

Use the same prompts, engines, market, settings, and collection window for every brand.

No benchmark invented

The result describes this sample only. It does not label a percentage good, bad, or typical.

Calculator

Calculate your sampled share of model

Calculate your sampled share of model

Sample inputs

Count brand and competitor mentions in one controlled sample.

Use the same prompts, answer engines, market, and collection window for every brand. Count each brand no more than once per answer.

Your sampled result

30.0%of counted category mentions belong to your brand in this sample.
Brand mentions12
Competitor mentions28
Denominator40
Your brand30.0%
Named competitors combined70.0%
Formula used12 brand mentions ÷ 40 total category mentions × 100 = 30.0%.
Discuss share-of-model measurement

Transparent measurement protocol

Freeze the evidence rules before collection, retain a prompt-level record, and compare only like-for-like runs. Inputs are capped at 100,000 answers; every brand count is also capped at answers reviewed. The workbook exports the normalized sample plus a documentation and QA template.

Protocol v1.0.0
  1. 1. Freeze the sampleRecord prompts, engines and model versions, market, language, settings, and collection window.
  2. 2. Apply one count ruleCount each named brand no more than once per answer, enforce the documented caps, and retain the supporting answer evidence.
  3. 3. QA the denominatorResolve ambiguous mentions, disclose missing data, and report no percentage when the denominator is zero.
  4. 4. Compare like with likeVersion protocol changes and avoid attributing movement to one cause without supporting evidence.

Methodology and caveats: This calculator reports only the sample entered here. It does not infer category demand, sentiment, citation quality, prompt coverage, or a market benchmark. The denominator is your brand mentions plus the three competitor mention counts. Because one answer can mention several brands, total category mentions may exceed answers reviewed. Inputs are capped at 100,000 answers; each brand count is capped at the smaller of answers reviewed or 100,000. If the denominator is zero, the calculator shows no percentage instead of treating absence as 0% share. Repeat the same documented prompt set over time to compare like with like.

A repeatable share-of-model tracking method

A repeatable share-of-model tracking method

1. Define the sample

Document category, prompts, engines, market, settings, and collection date.

2. Count consistently

Count each brand no more than once per answer, using the same competitor set.

3. Calculate the share

Divide brand mentions by all counted category mentions, then multiply by 100.

4. Repeat like for like

Rerun the same sample on a fixed cadence and investigate changes before drawing conclusions.

Share of model FAQ

Share of model FAQ

What is share of model?

Share of model is the percentage of counted category mentions assigned to a brand in a defined sample of AI-generated answers. It is a sampled visibility metric, not a universal market share.

How do you calculate share of model?

Add your brand mentions and competitor mentions to create the denominator. Divide your brand mentions by that total and multiply by 100. If the denominator is zero, report no percentage.

What should a share-of-model sample include?

Record the exact prompts, answer engines or models, market, language, date, settings, and counting rule. Keep them stable when comparing results over time.

Is there a good share-of-model benchmark?

Not from this calculator. A defensible benchmark depends on category, prompt set, engine mix, market, and sampling method. Compare like-for-like samples and disclose the method.