Recomma docs

Agent search

Where the search indexes an agent calls put your brand, which is not where the assistants put it.

Every other page under Measure reads a chat assistant: a question was asked, an answer came back, and the answer named somebody. An assistant that can search does not work that way. It calls a search index, gets a ranked list of URLs back, and writes from what it was handed — and the indexes it calls are not the ones a person uses.

This page is that layer, measured on the same questions as the rest.

Why both indexes

Two of those indexes can be measured directly, and they disagree.

Over thirty tracked questions across three brands, Brave and Exa found a brand eleven times between them and agreed on two of those eleven. A single index would have reported four fifths of this wrongly — in either direction. So both are drawn, and neither is averaged into the other.

These are two readings, not one score

There is no combined number on this page on purpose. Averaging an index that returns you at #1 with one that does not return you at all produces a figure that describes neither, and hides the only thing worth knowing: which of them an agent happened to call.

Indexes

The panel at the top carries two figures per index, and they have to be read together.

FigureWhat it answers
N of 20 questionsWhether the index hands your pages back when a buyer asks
holds at least N pagesWhether the index has your pages at all

The sentence above them says which case you are in, because the two combine into different problems with different fixes:

  • Not held — neither index has your pages, so no question can return them. That is a crawling problem before it is a content one, and Crawlability is where it starts.
  • Held and never returned — your pages are in the indexes and none of these questions reaches them. What the pages say is the thing to change, not whether they are there.

Coverage is a floor, not a total

Neither vendor sells a page count. Coverage is measured with a site: query, which returns one page of results and no total, so the figure reads “at least”. Both indexes reporting the same number usually means both hit that cap, not that they hold the same amount.

The questions

One row per question, ordered by where you are missing rather than by where you win. Questions no index returns you for come first, then the ones only one does, then the places that could be better. A brand that wins sixteen of twenty and one that wins two are both read from the top of the same table.

  • Where it ranks — a chip per index, carrying its name and your place, or a dash. Every index appears on every row including the ones that returned nothing, because a row reading Brave — beside Exa #2 is the disagreement this page exists to report.
  • Who is there instead — the hosts that came back where you did not, for that question. A row of dashes says you are missing; the same row with three domains on it says who to read to find out why.

Open a row for the page an index actually returned: ranked as it came back, your own result tinted, a tracked rival marked. The row is a summary of that page, so the page is what it opens.

How often it runs

One project is checked each scheduler tick and each project is checked weekly, so a newly tracked brand waits its turn rather than being checked immediately. Projects are ordered by age — oldest first — so a queue, not a draw.

If the deployment has no index key configured the page says so rather than reporting zero of twenty, which is a finding it has not earned.