definition
Share of Model (SoM) is the percentage of times an AI engine mentions or recommends your brand when responding to queries relevant to your category. It is the AI-era equivalent of Share of Voice — but measured across language models rather than media channels.
In traditional marketing, Share of Voice (SoV) measures how much of the conversation in a market your brand owns across advertising, PR, and social media. Share of Model applies the same principle to AI-generated answers. If a user asks ChatGPT "What is the best CRM for startups?" and the response names five products, each product that appears has a share of that model's answer.
The difference is that SoM is not about paid impressions or media mentions — it reflects the model's learned associations, training data, and real-time retrieval results. This makes it harder to influence through spending alone and more dependent on genuine brand authority, content quality, and technical AEO implementation.
calculation
How to Calculate Share of Model
Calculating SoM requires a systematic approach. The process involves defining your query bank, running those queries across AI engines, and analysing the responses for brand mentions.
[01] Define your query bank. Build a set of 50-200 queries that represent how your target audience asks AI engines about your category. Include informational queries ("What is the best..."), comparative queries ("X vs Y"), and recommendation queries ("Which tool should I use for..."). The query bank should reflect real user behaviour, not just your keyword list.
[02] Run queries across engines. Execute each query on ChatGPT, Perplexity, Claude, Google AI Overviews, and any other relevant AI engine. Use fresh sessions without prior context to avoid personalisation bias. Record the full response for each query on each engine.
[03] Extract brand mentions. Parse each response for mentions of your brand and competitor brands. Note the position of the mention (first, second, third), the context (positive, neutral, negative), and whether a citation link is included.
[04] Calculate the percentage. SoM = (number of responses mentioning your brand / total number of responses) x 100. Calculate this per engine and as an aggregate. Compare your SoM to each competitor to understand relative visibility.
cross_engine_variance
Cross-Engine Variance: Why One Score Is Not Enough
One of the most important findings in AEO measurement is that SoM varies dramatically across engines. A brand might have 45% SoM on ChatGPT but only 12% on Perplexity. This variance exists because each engine uses different training data, different retrieval mechanisms, and different answer synthesis approaches.
ChatGPT relies heavily on its parametric knowledge — what it learned during training — supplemented by web browsing when enabled. Perplexity, by contrast, performs real-time web searches for every query, making it more sensitive to recent content and less dependent on training data.
Google AI Overviews pulls from Google's search index, meaning traditional SEO signals have more influence. Claude draws on its training data with a different emphasis on source quality. Each engine requires a different optimisation approach, and measuring SoM per engine reveals exactly where your gaps are.
The practical implication: never report a single aggregate SoM number without the per-engine breakdown. The aggregate hides the variance that drives your strategy.
benchmarking
Benchmarking Against Competitors
SoM in isolation is less useful than SoM relative to competitors. A 25% SoM might seem low until you discover the category leader has 30% and the average is 8%. Competitive benchmarking transforms SoM from a number into an actionable insight.
To benchmark effectively, identify your top 5-10 competitors and run the same query bank for all of them simultaneously. Map the competitive landscape per engine: who leads on ChatGPT, who dominates Perplexity, and where the gaps exist. This competitive map reveals both threats and opportunities.
Track SoM monthly. Unlike SEO rankings that fluctuate daily, SoM tends to shift more gradually — especially on engines that rely on parametric knowledge. A sustained effort in content quality, technical AEO, and third-party citations typically takes 60-90 days to show measurable SoM improvement.
The most valuable benchmark is SoM on high-intent queries specifically. Being mentioned when someone asks "What is X?" is less valuable than being recommended when someone asks "Which X should I buy?" Weight your query bank accordingly.
beyond_frequency
Beyond Mention Frequency: Quality Dimensions of SoM
Mention position. Being the first brand named in an AI response is significantly more valuable than being the fifth. First-position mentions receive disproportionate attention and are more likely to influence user behaviour. Track not just whether you are mentioned, but where.
Sentiment accuracy. A mention is only valuable if the AI describes your brand correctly. If ChatGPT mentions your product but attributes incorrect pricing, outdated features, or a competitor's value proposition to you, the mention is harmful. Sentiment accuracy tracks whether AI descriptions match your actual positioning.
Citation inclusion. On engines like Perplexity that provide source links, being cited (linked to) is more valuable than being mentioned without a link. Citation inclusion creates a direct path from AI answer to your site. Track citation rate as a subset of mention rate.
Recommendation strength. There is a difference between being listed as an option and being explicitly recommended. "Consider X" is weaker than "X is the best choice for this use case." Tracking recommendation strength adds a qualitative dimension to SoM that raw mention counting misses.
operationalising
Operationalising SoM in Your AEO Programme
SoM is most powerful when it becomes the central KPI of your AEO programme rather than a vanity metric reported quarterly. To operationalise it, establish a baseline measurement, set quarterly targets, and connect every AEO activity to its expected SoM impact.
Build a SoM dashboard that breaks down performance by engine, by query category, and by competitor. Update it monthly. Use the data to decide where to invest: if your SoM on Perplexity is low because you lack recent, citable content, prioritise publishing. If your ChatGPT SoM is low despite good content, investigate whether your technical AEO implementation is blocking crawlers.
Finally, connect SoM to business outcomes. Correlate SoM improvements with changes in direct-to-site traffic, branded search volume, and pipeline metrics. As AI-mediated discovery grows, SoM will become as essential to marketing reporting as Share of Voice was in the media-driven era. The brands that measure it now will have the data advantage when the rest of the market catches up.
related_posts
What is AEO? A Complete Guide to AI Engine Optimization
AI Engine Optimisation is the practice of optimising your brand's digital presence for AI-generated answers. This comprehensive guide covers what AEO is, how it differs from SEO, and why every brand needs an AEO strategy.
Building Your Citation Network: How AI Decides What to Recommend
AI engines don't recommend brands randomly — they follow citation patterns across trusted sources. Learn how to build the citation network that gets your brand into AI responses.
ChatGPT vs Perplexity: Where Should Brands Focus AEO?
Two very different AI engines, two different approaches to brand visibility. A detailed comparison of how brands appear in ChatGPT vs Perplexity and where to invest your AEO efforts.