# What the Evidence page shows Updated: 1 Oct 2026. Source: https://www.botsay.ai/docs/evidence **Evidence** shows what the AI engines said about you, in their own words, and which sites and pages they cited. The figures on the other screens are counted from these answers. ## What are the two tabs of Evidence? Evidence has two tabs, each with a count beside its name. - **What they said**, with a count such as **221 answers named you**: the answers that named you, grouped by question. - **What they cited**, with a count such as **1,018 links**: every link the answers cited, grouped by site or page. Each tab opens onto detail. From **What they said** you can open a single answer, and from **What they cited** you can open a single cited page. The edition line at the top says which week you are reading and whether that week is still being read. ## What does the What they said tab show? **What they said** opens with a strip of four counts for the week, then a list of questions. The scope is stated at the foot of the page: **Counted over the questions that do not name you, the same set as your mention rate.** If no brand is marked as yours, the tab says **No brand is marked as yours**, with the link **Mark your brand in Setup**, because there is nobody for an answer to have named. If your questions have not been asked yet it says **Your first reading is not in yet**. If nothing in the current view named you, it says **No answer named you here** and suggests widening the list. ## What do the four counts at the top of What they said mean? The strip counts readings, where a reading is one question on one engine. A question asked more than once on an engine in a week is still one reading. - **Named you**: the readings whose answer named you, out of all the answers, for example "of 228 answers". - **Named you first**: of those, the ones that named you first, with the share, for example "40% of those". - **Described you well**: readings with a Sentiment of 80 or more. - **Described you poorly**: readings with a Sentiment under 40. The note reads **read these first** when there are any. Sentiment is BotSay's own estimate, as [how Sentiment is scored](https://www.botsay.ai/docs/the-five-measures#what-is-sentiment) explains, so the strip says estimated once instead of on each row. ## How is the What they said list ordered? Questions are ordered by the lowest Sentiment among the answers that named you, so you reach the worst thing said first. The heading **Least favourable first** says it. An answer with no Sentiment reading sorts after every scored one and does not count as the worst. Where scores tie, the worse place in the answer comes first. The line under the heading, **Sort: least favourable. Show: every engine, every theme**, has two menus. Press **every engine** to pick one engine and **every theme** to pick one theme. Only engines and themes that have answers this week are offered. A list over 25 questions is paged, with **Previous** and **Next** and a count such as "1 to 25 of 140 questions". ## What does a row on What they said mean? Each row is one question, shown with a quote from the answer that represents it: the least favourable one that named you. A row has: - A line above the question: the theme and how many answers were like this, for example "Pricing. 3 answers like this". - The question itself. - The quote. - A line saying who said it: "Google AI Overviews. Named 3rd. Sentiment 48", or "No sentiment reading". - A **See all quotes** button, which opens that answer. The quote is the only thing coloured, and only when its Sentiment is under 40. ## What does a single answer show? A single answer is one engine's reply to one question on one day. Its heading names the engine and the date, for example **ChatGPT, 14 Sep 2026**. One line says how it treated you: **Named you 2nd of 5 brands. Sentiment 62, estimated.** or **Did not name you or anybody else.** Beside the text sit three cards: - **Brands, in the order named**, with your row marked. - **Pages it linked to**, with a count. Each page is labelled **Yours**, **Rival** or **Publisher**. A publisher's page can add **Names you**, **Does not name you** or **Could not read the page**. - **Searches the engine ran**: the web searches it made, or a note that it does not report them. **Previous answer** and **Next answer** step through answers to the same question on the same engine. Your brand is marked in the brand colour and rivals in grey. ## Why does a single answer sometimes show lines instead of the full text? BotSay does not keep the full text of an answer once it has read it, so the page often shows only the lines that named a brand. It says so: **The full text of this answer is not kept. These are the lines that named a brand, in the order they appear.** The lines are the sentences each brand was named in, with you and rivals marked. Where an answer named no brand and no text is kept, the page says **Nothing is kept for this answer**. Three other states stand in for the text: - **Google showed no AI Overview for this question**: there was no answer to read. - **This answer is waiting to be read**: it will be ready after the next reading. - **This answer did not complete**: there is nothing to read for it. ## What does the What they cited tab show? **What they cited** ranks every site the answers linked to, under the **Your pages cited** figure with its change or **Last week** note, as on **This week**. If BotSay has found a site that may be yours and is not listed, a card, **This is understated**, offers a **Decide** button. Below it, a table headed **Every site the answers linked to** shows a count such as "1,018 sites", with these controls: - Chips: **All**, **Yours**, **Rivals**, **Places you could be**. - An engine select: **Every engine** or one engine. - A **Sites** or **Pages** select, which regroups the table by site or by page. - An **Export** link, which downloads the citation figures. [Export your data as CSV](https://www.botsay.ai/docs/exports) says what the file holds. A list over 25 rows is paged with **Previous 25** and **Next 25**. ## What do the columns on What they cited mean? The table has five columns, and only cited links count: a page an engine read without pointing at it is not in the table. - **Site** (or **Page**): the site, or the page address without its scheme. - **Kind**: **Yours**, **Rival**, **Community**, **Publisher**, **Company**, **Official body**, **Reference**, **Review site** or **Other**. - **What it means for you**: the move that fits the site, such as **Write your own answer** for a rival's site, **Ask to be listed**, **Take part**, **Pitch them**, **Get listed** or **Claim your profile**. Some are links to Work, and [Find pages to get your brand onto](https://www.botsay.ai/docs/where-to-get-in) explains the moves. **Already names you** means a checked page of that site names you. - **Links**: how many links the answers cited to it. - **Share**: the site's links as a share of all the links cited. ## What does a single cited page show? A single cited page opens from a row and gives the page's title and address, with **Open the page** to visit it. - **Answers that leaned on this page**: the count, with its share, for example "12.3% of 228 answers, across six readings. All six engines have cited it." - A line chart, by reading, for up to the last eight readings. - **What to do**: one instruction, for example **Write your own answer. A rival owns this page.** For a publisher's page it may add what the answers say about you when they cite it. - **Questions whose answers cited it**, ten at a time, each with "Names you in 4 of 6." and an **Open** link. On a page covered by an own-page rule, a button, **Mark as yours** or **Mark as not yours**, records your decision for that path. Marking does not change **Your pages cited** yet.