> ## Documentation Index > Fetch the complete documentation index at: https://docs.trysight.ai/llms.txt > Use this file to discover all available pages before exploring further. # Search + AI > Google rankings and AI answers in one visibility score, with share of voice per engine, a keyword-by-engine matrix, the citation gap and brand entities. ## Overview **Search + AI** lives under **SEO Suite → Search + AI** (`/seo/search-ai`). It joins the Google data the suite keeps (rank checks and organic keywords) with the AI answers collected for your tracked prompts, so you can see in one place where a keyword shows up: on Google, in Google's AI Overview and AI Mode, and in ChatGPT, Gemini, Perplexity and Claude. The page never calls an outside provider. It reads what is already stored: the latest rank check per keyword, the AI answers to your tracked prompts, the Google AI Mode checks and the AI Overview readings taken during rank checks. Included from **Pro**, as part of the SEO overview. The AI columns need [tracked prompts](/ai-visibility/tracked-prompts) linked to keywords; the Google AI Mode and AI Overview columns fill in once those checks run. The line under the title shows the sample: "N AI answers across X of Y prompts in the last 30 days". The selector on the right switches the AI answer window between the **last 7, 30 or 90 days**; 30 is the default. ## The score The **Search + AI visibility** card is one number from 0 to 100. Four components feed it, each weighted by search volume across your keywords: | Component | Weight | What it measures for a keyword | | - | - | - | | **Google organic** | 50% | Expected clicks at your position as a share of ranking first, using the site's click-through curve. Zero when unranked. | | **AI answers** | 30% | How often AI answers to the keyword's linked prompts name you or link to you: the average of the mention rate and the citation rate. | | **Google AI Mode** | 10% | The share of linked prompts where Google AI Mode linked to you at the latest check. | | **Google AI Overview** | 10% | Whether the AI Overview on the keyword's results page cited you at the latest rank check. | How it adds up: * For each component, the keywords that have something to measure are averaged, weighted by search volume. Keywords without volume count for nothing, unless no keyword has volume, in which case each counts once. * A component with no data for any keyword drops out, and its weight is shared out to the rest in proportion. The card lists the weights actually applied, so you can see when one was moved. * The result is rounded to a whole number. The score is stored once a day at **04:30 UTC**, after that day's rank checks, and the card plots the last 90 days with the change over that span. A keyword's AI Overview counts only when the results page showed an AI Overview; a page without one has nothing to measure and is not a miss. ## Share of voice by engine The **Share of voice by engine** table compares you with the competitors tracked in AI visibility, one row per engine, with the sample size in the last column. A dash means nothing has been sampled yet, not a 0% share. | Engine | Measure | Sample | | - | - | - | | **Google organic** | Volume-weighted expected clicks from positions on the latest rank check. | keywords | | **Google AI Overview** | Keywords whose latest rank check showed an AI Overview, counted for each site it cites. | keywords | | **Google AI Mode** | Latest checks that cite each site. When AI Mode cites you, only your sources are stored, so competitors are undercounted on those prompts. | prompts | | **AI answers** (ChatGPT, Gemini, Perplexity, Claude) | Answers naming each brand. Competitors count only under the names the detector tracks. | answers | ## Keywords across engines The matrix has one row per keyword in your keyword workspace and one column per engine. | Column | Meaning | | - | - | | **Keyword** | The keyword, with how many prompts are linked to it and how many are still pending. Click it to open the keyword. | | **Volume** | Monthly search volume. | | **Google** | Your organic position on the latest rank check. | | **AI Overview** | Whether Google's AI Overview cites you for this keyword. | | **AI Mode** | Whether Google AI Mode links to you on the linked prompts. | | **ChatGPT**, **Gemini**, **Perplexity**, **Claude** | What each assistant did on the linked prompts. | | **Clicks** | Search Console clicks in the last 28 days. Off by default; turn it on from **Columns**. | Each AI cell shows one status: | Status | Meaning | | - | - | | **cited** | At least one answer linked to the site. | | **mentioned** | An answer named the brand without linking to it. | | **absent** | Answers came back and did neither. | | **pending** | The prompt is tracked but has no answer yet. Pending is never counted as a gap. | | dash | Nothing to measure: no prompts are linked to the keyword, or the engine has no data. | The **View** filter narrows the table: * **Ranking on Google, absent in AI**: on page one (top 10) but no AI engine names or cites you. * **Cited in AI, not ranking**: cited somewhere in AI while outside the top 10. * **Absent everywhere**: unranked and absent in AI. * **No prompts linked**: keywords with nothing tracked on the AI side. Link prompts to a keyword from its drawer under **Keywords**. Answers from Grok count toward the AI answers rates even though Grok has no column of its own. The toolbar has a search box, a **Columns** menu and a CSV export. ## Citation gap by source type This card looks at prompts that were answered in the window and never cited you, and groups what the answers cited instead: **Competitors**, **Review sites**, **Reddit**, **YouTube**, **News**, **Docs and help centers** and **Other sites**. For each group you see how many prompts and citations it accounts for, the top three domains, and one action: | Group | Action | | - | - | | Competitors | **Write comparison article**: opens the agent with a "you vs competitor" keyword. | | Review sites | **Get listed via Outreach**: opens the Outreach inbox. | | News, YouTube | **Pitch via Outreach**: opens the Outreach inbox. | | Reddit, Docs and help centers, Other sites | **Brief**: copies a short plan with the most cited domains and example prompts. | Your own pages are ignored when grouping. The card reads "No gap yet" when every answered prompt cites you, or when nothing has been answered. ## Brand entities AI answers do not always use your exact brand name. **Brand entities** holds the other names they use: acronyms, product names, handles. Add up to 25, then **Save**. The names live with the site's AI visibility brand settings, so AI visibility must be set up for the site first. The next daily score re-reads up to 2,000 recent answers that the detector did not count as a mention and checks them for the names. Matches count as mentions in the score and the matrix, and the card shows how many extra mentions were found so far. Stored detections are not rewritten. Start with the **Ranking on Google, absent in AI** view. Those keywords already have the page and the ranking; the citation gap card tells you which kind of source the AI answers trust instead. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.