Sources and citations

When an AI model answers a question, it usually draws from specific web pages to build that answer. AI Peekaboo captures exactly which pages, and which brand each one was actually discussing, so you can see what's feeding the AI's answers about your category.

How attribution works, per model

Each AI model exposes its sources differently, so Peekaboo reads each one using its native citation format:

• ChatGPT marks specific spans of text with citation annotations showing exactly which URL backs that span. That span of text is checked for brand names: a span with one clear brand match gets high-confidence attribution, a span mentioning multiple brands gets a lower confidence and defers to whichever brand seems primary, and a span with no clear brand match is stored as an unattributed source.
• Gemini provides its own grounding metadata, linking specific source URLs directly to the specific parts of the answer they backed, each with its own model-reported confidence.
• Perplexity is read through its native citation system along with its inline numbered references in the answer text.

Every captured source is stored with its URL, page title, domain, an estimated domain authority, and a content category, matched against a classification list, for example Community Forum, News/Media, Tech Publication, Review Site, or Corporate Official. This includes sources like Reddit threads, YouTube videos, Trustpilot reviews, and other community or review platforms that AI models frequently cite, not just brand websites.

Where to see it

The Sources page has two views:

• Domains: a ranked list of which domains get cited most often, broken down by content type, with a heatmap showing which AI models cite which domains.
• URLs: the specific pages that got cited, how many times, and when each was last seen in an AI answer.

You can search, filter by time range, model, or topic, and drill into the specific prompt or brand a source was attributed to.

Top Sources

Your Top Sources ranking sorts domains by how many times they were cited, along with each domain's percentage share of total citations in your selected window. Source Types shows the same ranking grouped by content category instead of by individual domain.

Exporting

Sources data can be exported as CSV, including reference counts, sentiment split, and authority data, so you can bring it into your own reporting or hand it to a content or outreach team.