What is Share of Model?

How does Share of Model work?

Share of model narrows a visibility analysis to a particular model or assistant environment. A practical definition uses a brand’s qualifying appearances divided by all appearances for a selected competitor set.

Other providers may use recommendation weights or response-level inclusion instead.

Specify which interpretation applies before reporting a number. An API model with web search enabled is not equivalent to the same model without retrieval.

A consumer ChatGPT session can also differ from an API request because its configuration and available context are not necessarily identical.

Why does Share of Model matter for B2B SaaS?

B2B SaaS buyers may rely on different assistants for research. An aggregate visibility number can conceal strong presence in one environment and weak presence in another.

A model-specific view helps identify where to inspect missing evidence, incorrect positioning, or inconsistent brand descriptions.

The metric should not be described as the model’s internal preference or memory. Observed outputs cannot reveal a hidden brand score or prove that training data changed.

They only describe the responses collected under the documented conditions and sampling design.

What is an example of Share of Model?

Imagine a hypothetical audit of a named model with web retrieval enabled. One fictional vendor receives fifteen of sixty competitor appearances, producing a twenty-five percent appearance share under the declared method.

The analyst preserves prompts and outputs for review.

A second run without web access may produce a different result. That difference does not isolate the cause because retrieval availability changes the task.

The report should label both configurations rather than merging them into a single supposedly inherent model preference.

How do you improve or measure Share of Model?

• Record the model identifier when available and name the assistant interface used.
• Document retrieval settings, prompt wording, context, and collection dates.
• Keep competitor definitions and visibility units consistent across runs.
• Repeat prompts to distinguish persistent patterns from variable answers.

Compare model-specific presence with citation accuracy and qualified site visits. Treat a change as a signal to investigate rather than a proven response to an optimization.

Official engine documentation explains features, but it does not define a shared industry standard called share of model.

Frequently asked questions about Share of Model