What is share of model?

Share of model is the fraction of answers an AI engine gives to a category's buying questions that name your product. Ask ChatGPT, Perplexity, Gemini, Grok, and Google AI "what should I use for social listening" every day for a month, count how many answers include you, and that fraction is your share of model. Tom Roach gave the metric its name in a Marketing Week column arguing that brands now need to be known by models, not only by people.

Why share of model matters for a founder

A growing slice of buyers never open a search results page. They ask an assistant, read the shortlist, and try the first two names. If you are not on that shortlist, you lose a buyer you never saw, and no dashboard records the loss. Share of model is the only number that makes that loss visible. It is also a leading indicator. A product that appears in one answer out of ten today tends to appear in more once its category threads improve, and a product that drops out usually dropped out of the cited sources first.

How to measure share of model

Start with a prompt set: the ten to thirty questions a buyer in your category actually types. Ask each question to each engine on a fixed schedule, since answers vary run to run and a single ask tells you little. Record three things per answer: whether you were named, who was named instead, and which pages or threads the engine cited. The naming rate is your share of model. The other two are your to-do list. Oetzi runs this check on a schedule for two credits per question across every engine and lists the cited threads next to the score.

What moves share of model

Engines compress what other people wrote, and they discount a vendor's own site heavily. What carries weight is the Reddit thread, the Hacker News comment, the comparison post, and the documentation page that the engine retrieves when it answers. The practice of improving those sources is generative engine optimization. The shortest path is to show up, disclosed and helpful, in the threads the engines already cite, and to fix the stale comparison that names your old pricing. There is a longer introduction in the share of model post.

Questions

Is share of model the same as share of voice?

No. Share of voice measures how much of a category's paid or earned media mentions you, and it is measured across humans and publications. Share of model measures how often AI engines name you when asked a buying question. The two move on different inputs. You can have loud ads and zero presence in model answers, or the reverse.

How often should I check share of model?

Weekly is enough for most products. Engine answers shift when models refresh, when a new thread ranks, or when a competitor launches, and none of those happen daily. A weekly schedule across five engines gives you a trend line within a month while keeping the cost low. Check daily only around a launch or a pricing change.

Can I raise share of model quickly?

Not reliably, and anyone promising a fixed timeline is guessing. What you can do quickly is answer the live threads in your category, correct the comparisons that engines cite, and publish the plain page that answers a buying question directly. Those changes show up in retrieval based engines within weeks and in model memory over a longer horizon.