Share of model is a simple idea with a growing payoff: across the questions your buyers ask AI, how often does the model mention or recommend you versus your competitors? It is the AI-era descendant of share of voice and share of search, moved into the place where buyers increasingly form their shortlist, inside the model’s answer.
I’m Andrii Byzov, a fractional CMO for B2B tech. Marketing teams have always tracked some version of presence: share of voice in advertising, share of search in demand. As research shifts into AI assistants, a new version is needed, because being ranked fourth on Google means little if the model never names you. This is what share of model is and how to track it honestly.
Key takeaways
- Share of model is your share of mentions and recommendations inside AI answers versus competitors.
- It is the AI-era version of share of voice and share of search.
- Being named in an AI answer is the new being shortlisted.
- Measure it with a fixed prompt set, run across engines on a schedule.
- Treat it as directional: repeat runs, report ranges, watch the trend.
Why a new metric is needed
Forrester has argued that AI search cracks the foundation of B2B marketing’s accountability model. Rank tracking and organic-session attribution weaken when the answer is synthesized inside a model and no click follows. Teams still need to answer a basic question for the board: are we gaining or losing presence where buyers now look? Share of model is a direct way to answer it.
How to measure share of model
- Define the prompts. Pick a stable set of buyer-intent questions: category questions, comparison questions, and pain-point questions.
- Run across engines. Execute them on ChatGPT, Perplexity, Gemini, and Google AI Overviews on a regular schedule.
- Record mentions. Note when you and each competitor are named or cited.
- Compute the ratio. Your mentions divided by total brand mentions is your share of model.
- Repeat and trend. Outputs vary between runs, so repeat and report a range over time rather than a single number.
This is the measurement layer behind AI visibility, and it pairs with the growth work in AI search optimization.
What to do with the number
A low or falling share of model is a prompt to act, not panic. The levers are the same ones that move AI presence generally: a clear entity, structured content, answer-shaped pages, and trusted citations. If the model describes you inaccurately rather than just rarely, that is a AI reputation management problem, which is a different fix.
The honest caveats
Share of model is directional. Model outputs shift between runs and between updates, so a single snapshot is noisy. Methodology has to be transparent: same prompts, same cadence, repeated runs, reported as a trend. Used that way it improves your odds of investing in the right AI-presence work. Treated as a precise score, it will mislead you.
For the wider context, see how to measure AI marketing ROI and what an AI-native fractional CMO does.
FAQ
What is share of model? The proportion of AI answers, across a set of buyer questions, in which your brand is mentioned or recommended versus competitors. It is the AI-era cousin of share of voice and share of search.
How do you measure share of model? Pick a fixed set of buyer prompts, run them across ChatGPT, Perplexity, Gemini, and AI Overviews on a schedule, record mentions for you and competitors, and track the ratio over time using repeated runs.
Why does share of model matter? As buyers ask AI which tools to consider, being named is the new being shortlisted. A higher competitor share of model means they enter more consideration sets than you.
Is share of model a reliable metric? It is directional, not exact. Measured consistently with repeated prompts over time, it gives a useful trend, but a single measurement is noisy.