Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minuteShare of model (SoM) is an emerging metric for how often a brand appears in AI-generated answers to a defined set of relevant prompts. It can help teams track visibility in AI search, but there is no single industry-standard formula. A SoM percentage only means something when its sample, counting rule and denominator are disclosed.
What share of model measures
SoM adapts the idea of share of voice to AI answers: instead of measuring a brand’s visibility through a channel such as search rankings or advertising impressions, it counts a defined kind of presence in answers from selected AI systems. A current glossary describes the metric as AI visibility, while measurement guides use different operational definitions; none establishes a formal industry standard. CDP’s glossary and Riseklix AI’s measurement guide are examples of those approaches.
Most importantly, SoM is bounded by the prompts and systems actually tested. A result for a particular question set, platform, language, geography and time period is not automatically a measure of visibility across all AI answers, nor is it a measure of market share.
Which kind of visibility are you counting?
Before calculating SoM, decide what qualifies as a brand appearance. These signals answer different questions and should be reported separately.
#1 Best Overall
- Brand mention: Does the answer name the brand at all? A passing reference counts under a presence-based rule, even if the answer does not recommend it.
- Recommendation: Does the answer suggest the brand as an option, or identify it as a first choice? This is a narrower signal than any mention.
- Source citation: Does the answer cite a domain? A cited source is not the same thing as a brand mention: a brand can be named without its site being cited, and a site can be cited without the brand being recommended.
Do not combine these into one unlabeled score. A glossary that distinguishes mentions, recommendations and citations is CDP’s September 28, 2026 overview.
Two formulas that produce different results
There is no canonical denominator. State the one used beside every reported percentage; otherwise, readers cannot tell what the number represents.
Rank #2
| Metric | Example calculation | What it says |
|---|---|---|
| Answer-level mention rate | Eligible successful answers naming the brand ÷ all eligible successful answers × 100 | The share of eligible answers in the sample that contain the brand. One answer may name several brands, so brand rates can add up to more than 100%. |
| Share of tracked brand mentions | Mentions of the brand ÷ all tracked brand mentions × 100 | The brand’s portion of the mentions counted across the sample. This is not the same as the percentage of answers that mention it. |
The first formula is an operational approach described by Riseklix AI, not a universal standard. A separate guide illustrates the arithmetic with a brand appearing in 30 of 100 representative prompts: that is a 30% inclusion rate for that particular sample, not a market benchmark or an independently measured industry statistic. AIO Copilot’s June 21, 2026 guide presents the example.
How to measure AI answer visibility consistently
- Define the question set. Build a fixed bank of realistic prompts related to the category and the audience’s tasks. Document the exact queries and the rule for including an answer. Keep the raw answers or results needed to audit the count.
- Specify what counts. Choose whether you are tracking mentions, recommendations, first choices or citations, and decide how to handle answers that include multiple brands. Record the numerator and denominator explicitly.
- Record the systems and conditions. Name the AI assistants or answer surfaces tested, whether browsing or retrieval is enabled, and the language, geography and audience context. Results from different platforms or modes should not be silently blended.
- Set handling rules for incomplete runs. Record collection dates and repetitions. Define how you treat failed responses, refusals and other ineligible answers, and make that rule visible alongside the denominator.
- Repeat the same design to track change. Rerun the same prompt panel on a stated cadence and retain the raw counts. AI responses can vary between runs, so repeated measurement makes a trend more interpretable than a single snapshot, but it does not remove that variability. SEOforAI.net’s July 11, 2026 overview notes this directional limitation.
For useful segmentation, report results by intent, market or platform when the sample supports it. A blended figure can hide that a brand appears frequently on one assistant and rarely on another.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteHow to compare two SoM reports
Two dashboards may both label a result “share of model” while measuring different things. Before comparing their percentages, check whether they use the same:
- Unit: answer-level brand presence, recommendation or first choice, or source citation.
- Denominator: eligible answers, all tracked brand mentions or all citations.
- Prompt panel: the same queries, buying tasks and eligibility rules.
- Platform and mode: the same assistants or answer surfaces, with browsing or retrieval conditions disclosed.
- Market and timing: language, geography, audience, category and collection period.
- Sampling method: number of runs, cadence, and treatment of failures or refusals.
- Competitive set and sentiment: which brands are counted and whether neutral or negative mentions are included.
Ask for the query bank, inclusion rules and raw counts if a report does not show them. Without those details, a difference between two percentages may reflect different measurement choices rather than a change in visibility. Riseklix AI’s guide specifically recommends checking these disclosures when evaluating dashboards.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What a SoM result can—and cannot—tell you
SoM is best treated as a directional visibility measure under a stated methodology. A higher mention rate does not, by itself, demonstrate that users prefer the brand, that an answer is accurate or trustworthy, that people click through, or that revenue or market share has increased. A citation rate answers a different question from a brand-mention rate, and neither alone establishes business impact.
AI visibility tracking software may help collect and organize answers across a prompt sample. When assessing a tool, check whether it exposes the prompts, systems and modes tested, its counting definitions, raw results and repeat cadence. The existence of this software category does not establish that any one vendor’s metric is authoritative. Cited’s overview discusses the category; it is not an independent validation of a universal SoM standard.
Free tools Windows power users keep installed
One-click scans. No signup required.
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.
