Upcoming EventSemrush Spotlight London
heyoo.ai

Share of Model

AI Search Visibility

Share of model is the share of AI-generated answers, across a defined set of prompts and a defined set of AI tools, in which a brand, product or person gets mentioned or cited. It’s built the same way share of voice is built for traditional media or search: pick a representative basket of questions your buyers would realistically ask, run them, and count who shows up.

The difference from share of voice is what’s being measured inside. Share of voice typically counts media mentions or search rankings. Share of model counts appearances inside a generated answer, which can be a citation with a link, a quote without one, or simply a mention with no source at all.

Key takeaways

  • Share of model is measured the same way share of voice is: define a basket of realistic buyer prompts, run them across target AI models, and count how often each brand appears relative to competitors.
  • Share of model can vary meaningfully between AI engines. A brand can lead on Perplexity and barely appear on ChatGPT for the same prompt set, because each engine retrieves and weighs sources differently.
  • Because model outputs shift as models are updated and retrieval indexes refresh, share of model needs to be tracked on a recurring cadence, not measured once as a benchmark.

What is share of model?

Share of model answers a specific question: out of every time a relevant question gets asked to an AI model, how often does this brand show up in the answer, compared with its competitors. It’s typically expressed as a percentage: if a brand appears in 18 of 100 prompt runs across a category, and a competitor appears in 40, the competitor has more than double the share of model.

The metric only means something relative to a specific prompt set and a specific set of models. A brand’s share of model for 'best employee advocacy platforms' and for 'how to measure LinkedIn engagement' will differ, sometimes sharply, because the underlying source material for each question is different.

How is share of model different from share of voice?

Share of voice traditionally measures presence across a defined media or search landscape, PR mentions, ad impressions, or search rankings, against competitors. Share of model measures the same competitive question, but inside AI-generated answers specifically.

Two practical differences matter:

  • Volatility. Search rankings move gradually; AI model outputs can shift answer to answer, and shift more sharply when a model is updated.
  • Engine fragmentation. Share of voice is often measured against one dominant channel (Google). Share of model has to be measured separately across several materially different engines, since performance on one doesn’t predict performance on another.

How do you measure share of model?

The core method: assemble a basket of prompts a real buyer would ask at each stage of research (category questions, comparison questions, 'best X for Y' questions), run them across target AI tools on a recurring schedule, and log every brand mention against the total.

Dedicated AI-visibility platforms, such as Profound, automate this at scale and track it over time. Smaller teams can start with a manual version: a fixed prompt list run monthly across two or three tools, logged in a spreadsheet. The manual version is directionally useful even without automation, as long as the prompt set and cadence stay consistent.

Activate your team on LinkedIn

Heyoo helps marketing teams turn employees into authentic, on-brand storytellers, with personalised drafts, a shared calendar, and pipeline-grade analytics.

Frequently asked questions

How often should share of model be measured?

Monthly at minimum, given how often AI models update their retrieval behaviour and underlying weights. Categories with fast-moving competitive dynamics may warrant a tighter cadence.

Does share of model replace share of voice?

No, it complements it. Share of voice still counts for traditional search and media presence. Share of model measures a newer, separate surface, AI-generated answers, that increasingly influences buyer research before a traditional search ever happens.

Can share of model be measured without paid tools?

Yes, with more manual effort. A fixed set of prompts run by hand across two or three AI tools on a consistent schedule produces a directionally useful, if more labour-intensive, version of the same metric that dedicated platforms automate.

Related terms