Skip to main content
The first tracking run starts once your workspace has a brand, a website and prompts. A single run gives you a reading; a trend needs several, so give it a few days before concluding anything from a movement.
Because the assistants are generative. The same question produces different wording, different ordering and sometimes different recommendations each time. That variance is why prompts are asked repeatedly, and why one run is a reading rather than a measurement.
No. Surfaced measures whether you are recommended and cited, not whether anyone clicked. Click data lives in your own analytics, and the two answer different questions.
No, and it could not. Surfaced asks its own questions, the prompts you configure, as an anonymous user. Nothing about real users’ conversations is visible to it or to us.
Because it is a cited source like any other, and seeing which of your pages get quoted is half the value of the screen. It is highlighted rather than removed, and it counts toward the totals.
A gap is a finding: something the answers get wrong or leave out. An action is work: something a person is going to do about it, with an owner and a state. Accepting a suggestion on a gap is how one becomes the other.
No. It records that you did something. Whether the answers changed is what the next tracking run tells you, and expect a lag of at least one run, usually more, because the engines have to re-read your site first.
A dash means nothing was measured, not that you scored zero. If no brand at all was detected in the period, there is no denominator. A real zero, where competitors were named and you were not, is shown as 0% with that wording.
Enough to cover the questions that decide a purchase, which for most businesses is a few dozen rather than hundreds. Prompts you add for completeness dilute your visibility figure without telling you anything. Start with the questions your sales team gets.
A few, to see how you are described when someone already knows you. Keep them separate in your reading: branded prompts return your brand nearly always, so blending them makes your visibility look far healthier than it is. More.
Yes, one workspace per brand. Workspaces have their own prompts, competitors, members and settings, and sit under one organization for billing. More.
Prompts above the cap are paused rather than deleted, with a banner saying how many, and history older than your plan’s window stops being reachable rather than being erased. Raising the plan brings both back. More.
Analytics exports as a PDF carrying your selected period, and gap rows export to CSV from the bulk bar. There is also a public API for programmatic access.
It reads your homepage and your priority pages rather than crawling everything, which keeps a re-check to minutes and keeps the score focused on the pages that get cited. It is a sample, and the header tells you how many pages it covered.