The country we ask from and the AI engines we ask. Each engine is asked every question.
The three defaults
ChatGPT, Google AI Overview and Perplexity collect for every project, on every plan, from the day the project is created.Opt-ins
Gemini, Google AI Mode and Claude Haiku are on every plan and off until you switch them on for a project.- Switching one on needs a paid subscription. During the trial the row reads “On your plan once your subscription is paid”.
- Only an owner or admin can switch one on. An editor can keep one that is already on, or switch it off.
- A workspace can have up to six extra engines switched on at once, across all its projects.
- An extra engine collects from the next daily collection after you save.
- An extra engine’s answers are collected and kept, but no screen counts or lists them: every figure and chart covers the default engines you have on. Its answers and figures are available only through the API’s
enginesfilter.
Enterprise only
Claude (Sonnet), Grok and Claude Opus are available on Enterprise plans only. Project settings lists them with “Enterprise only”. Contact us to talk about Enterprise.Not tracked
Microsoft Copilot is not tracked. Its answers can only be read signed in, and we collect every answer signed out. In Attribution, visits that come from Copilot, Claude or Grok are counted and labelled “not sampled” rather than dropped.Switching an engine on or off
1
Open Project settings
Open the project and choose Project settings in the rail.
2
Go to Market & engines
Each engine has a switch and one line on its state, such as “Running your prompts”, or why it cannot be switched on.
3
Save
Save the settings. The change applies from the next daily collection.
The market
Next to the engines, Project settings shows the project’s market: “Set when your prompts were created. Prompts run from this market’s location; a second market is a second project.” Today every project asks from the United States, in English.Related
Project settings
Everything else on the screen the switches are on.
Plans and prompt limits
What each plan includes.
How answers are collected
How each engine’s answer reaches us.
What we do not measure
The engines and answers outside our sample.