AI answers

Performance → AI answers shows how AI engines answer the questions your site could be cited for and who they cite instead. Crawler traffic has its own Performance → Crawlers tab.

Engines and questions

SERPclimber checks Google AI Overviews, ChatGPT, Perplexity, Gemini, and Claude. The question universe is built from the queries you already rank for in Search Console, so it needs no setup: once Search Console is connected and queries are available, the first scan starts on its own. Questions are added in batches and each engine's answer is checked for a citation of your site, a competitor, or neither.

A daily job keeps saved answers fresh. Choose Run scan to check the next batch of questions now; it reads Scanning… while it runs and can be limited by the project's monthly budget. Scans use the project's paid evidence budget, never your live-question allowance. If the budget is exhausted, saved results stay visible and the scan resumes next month.

Cited includes both a source citation of your page and a link to your site in the answer. For example, an answer can use another site's advice and still recommend your tool with a link; that counts toward your Share of answers. The answer's sources distinguish Source citation from Answer link. Answer links have no source rank. A brand name without a link does not count as cited.

Read-only projects cannot scan; the page offers Reactivate it in Billing.

Reading the page

The toolbar has the same Filter, metric, and date-range controls as the other Performance tabs. Filters are engine, status (cited, competitor cited, not cited, no answer), change (gained or lost in the last 30 days), page, and question text.

The headline is Share of answers: of every AI answer to your tracked questions (one answer per question and engine), the percentage that cites your site, with its change in points against the comparison window. Its chart plots that share per day, the previous period ghosted beneath it, with a flag on the day an engine first cited you. The band above the line shows where the rest of the answers went: competitor share (amber — an engine cited a rival instead of you) and uncovered share (grey — nobody was cited). Snapshots taken before this split was recorded still show your share, but not who else was cited. If snapshots begin after the start of the selected range, a quiet note marks the first day with data.

The tiles in the chart header — Share of answers, Competitor share, Uncovered share, and Opportunities (questions where a competitor is cited and you are not) — each jump to the section that explains them. Under the chart, one row per engine shows that engine's share, a sparkline of it over the range, and its change; click a row to filter the page to that engine (click again to clear it). Engines with fewer than three answers in the range show the count instead of a share.

Engine filters apply to the whole hero, because the daily history is recorded per engine. Question-text, status, change, and cited-domain filters cannot be read back from the history, so under those the tiles compute from the matching questions and the chart says so. Citations Gained and Lost live in the change filter and the Lost recently card.

The sections below each carry an Explain with AI button:

• Engine coverage — one bar per topic, biggest topics first. Topics are named by the analysis model at scan time from the tracked questions themselves (every question is assigned; the names stay stable between scans), so they read the way you would say them aloud. Each bar is drawn to the topic's size — its width is the topic's AI answers against the largest topic's — and splits those answers into the ones that cite you (green), the ones that cite a competitor instead (amber), and the ones that cite nobody (grey); the engine marks under it light up for engines that cite you at least once. The right column shows the share of the topic's answers that cite you, its change against the comparison window, and the underlying "n of m answers". Topics with a single question fold into one "Smaller topics" row, and questions that fit no topic sit in "Other questions"; both always close the list and are never cut off by the twelve-row limit. Click a segment to see those questions, an engine mark to filter the page to that engine and topic, or Expand to list every topic exhaustively in the same bar layout (each smaller topic on its own row, then the other questions) — clicking the "Smaller topics" label opens it at those rows — with the topic × engine grid as a second view.
• Competitors are winning these — questions where a rival is cited and you are not, ranked by demand and how many engines cite the rival. Each row shows the rival, the page of yours the engines should cite when one matches, and a Create task button that files an AI answer gap task for the engine.
• Lost recently — questions that dropped a citation in the range, with the engine, when it was lost, and who holds the spot now.
• What earns citations — your pages ranked by citations received across engines, with the number of questions each page answers and its strongest engine. Click a page to see those questions.
• Competitors — the domains engines cite in your space: how many questions each wins against you, on which engines, with a breakdown on click.

Question rows show: the question, a dot per engine (filled when you are cited, hollow when absent, ringed when a competitor is cited), and a demand bar. Open a row to read each engine's answer and its sources. Long questions in the Asked so far list wrap to fit the dialog. Ask the engines lives in the toolbar: one live question can check any engines you select and uses one of the 500 monthly questions each active site adds to the shared account pool. Each selected engine counts toward the site's 100-engine-check daily safety limit.

All tracked questions at the bottom is the full table: question, engines, best status, the page cited, demand, and when the question was discovered. Open a row to read each engine's answer and its sources.

From answers to work

Citation gaps and blocked crawlers become Audit findings under AI visibility, where you can create a task. In chat, ask the assistant to explain any card, or ask it directly what to write to win a specific question.