What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Share of model (SoM) is an emerging way to measure how often a brand appears in AI-generated answers to a defined set of relevant prompts. It resembles share of voice, but its result applies only to the prompts, AI systems, markets, and period measured—not to all AI answers or to market share. Because there is no settled industry formula, a useful SoM report must explain exactly what it counts and what it divides by.
What share of model measures
In AI search and answer experiences, share of model describes a brand’s visibility within a specified sample of responses. The sample might use questions about a product category, comparisons between providers, or other tasks a target audience might ask an AI assistant.
As an Amazon Associate I earn from qualifying purchases.
The term is analogous to share of voice, but the measurement channel is different. Traditional search visibility may be assessed through rankings or impressions; SoM looks inside AI-generated answers. The label alone does not tell you whether a report counts any brand mention, recommendations, or citations. Those are separate signals, and reports should name the one being measured. CDP.com’s definition distinguishes these measures, while emphasizing that SoM is an emerging metric rather than a formal standard.
Why the formula matters
There is no canonical SoM formula. Two commonly used approaches answer different questions:
#1 Best Overall
| Measure | Calculation | What it tells you |
|---|---|---|
| Answer-level mention rate | Eligible successful answers naming the brand ÷ all eligible successful answers × 100 | How often the brand appeared in the sampled answers. One answer can name several brands, so rates across brands need not add up to 100%. |
| Share of tracked brand mentions | Mentions of the brand ÷ all tracked brand mentions × 100 | What portion of the counted brand mentions belonged to that brand. |
The first is an operational formula described by Riseklix AI; it is not an industry-wide rule. A separate illustrative example from AIO Copilot says that a brand appearing in 30 of 100 representative prompts has a 30% inclusion rate for that sample. That is an arithmetic illustration, not a benchmark or a finding about the market.
Always report the numerator and denominator with the percentage. “30% SoM” is ambiguous unless readers know whether it means 30% of eligible answers mentioned the brand or 30% of all tracked mentions belonged to it.
Rank #2
Mention, recommendation, and citation are not interchangeable
- Mention: The answer names the brand, whether or not it endorses it.
- Recommendation: The answer proposes the brand as an option, perhaps as a shortlist choice or first choice. A passing reference does not necessarily qualify.
- Source citation: The answer cites a page or domain. This measures source visibility, not necessarily whether the brand itself is named or recommended.
Track these separately if they matter to your goal. A brand can be mentioned without being recommended, and a cited source can appear without a clear brand endorsement.
How to measure AI visibility with a repeatable sample
- Define the question bank. Choose realistic, category-relevant prompts and state which prompts qualify. Keep the actual list so later measurements can use the same questions.
- Specify the systems and settings. Record the AI assistants or answer surfaces tested, including whether browsing or retrieval is enabled. Results from different platforms or modes should not be blended without disclosure.
- Set the counting rules. Decide whether the unit is any answer naming a brand, an explicit recommendation, a first-choice recommendation, or a cited source. Define how to treat neutral and negative mentions, answers naming multiple brands, refusals, and failed runs.
- Record context and raw results. Keep the language, geography, audience or category, collection dates, eligible-answer count, and brand counts. Segment by intent, market, or platform when that helps explain meaningful differences.
- Repeat the same design. Re-run the fixed panel on a stated cadence and compare like with like. AI answers can vary between runs, so retain run-level counts rather than presenting one result as a definitive score.
For example, if a defined panel produces 100 eligible successful answers and 30 name a brand, the answer-level mention rate under that method is 30%. If the denominator instead includes every tracked brand mention, the resulting percentage answers a different question. AIO Copilot’s 30-of-100 example illustrates the first kind of inclusion calculation; it does not establish a universal sampling method.
How to compare two SoM reports
Before treating a change or competitor gap as meaningful, check whether both reports match on these dimensions:
- Unit and denominator: Answer-level presence, recommendations, or citations—and all eligible answers, all brand mentions, or all citations.
- Prompt set: The same queries, tasks, and eligibility rules.
- Platform and mode: The same assistants or answer surfaces, with the same browsing or retrieval conditions.
- Market context: Language, geography, audience, category, and measurement period.
- Sampling: Number of runs, cadence, and handling of refusals or failures.
- Competitive and sentiment rules: Which brands count, and whether neutral or negative appearances are included.
If those details differ, the figures may not be directly comparable. Riseklix AI advises readers to request the query bank, inclusion rules, and raw counts when assessing measurement dashboards. A vendor’s chosen formula can be useful, but it should not be mistaken for a settled standard.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What a share-of-model result can—and cannot—tell you
SoM can help track whether a brand appears more or less often in a consistent sample of AI answers. It is best treated as a directional visibility measure, not a complete measure of marketing performance. A percentage by itself does not establish that users prefer the brand, that an answer is accurate or trustworthy, or that visibility produced clicks, revenue, or market share. Nor does one blended score reveal whether visibility differs sharply by platform.
As SEOforAI.net notes, answers can vary, which is another reason to interpret repeated measurements as a trend under a stated method rather than as an exact, universal ranking. If you use tracking software, check whether it exposes its prompts, tested engines, counting definitions, raw results, and repeat cadence; a dashboard score without those details is difficult to evaluate.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
