Source types and ownership

You open the Sources page to decide where to earn citations, and every row carries labels: a type for the site, a type for the page, and an owner. They answer three different questions, and mixing them up leads to the wrong plan. A review directory is not always a competitor, and a competitor's blog post is still a competitor's page.

Three different labels

The same site can carry very different labels. A software company's website is Software & Tools by site type. If you sell a competing product and track that company, it is also Competitor by owner. One of its pages can be a Comparison page by page type.

Site Type: what kind of site

Every surface that names the kind of website a citation came from uses one shared list of 21 site types:

• Social Media
• Reference & Encyclopedia
• News & Media
• Software & Tools
• Marketplace
• Travel & Booking
• Dating & Relationships
• Reviews & Directories
• Jobs & Careers
• Guides & Comparisons
• Healthcare
• Banking & Finance
• Marketing Agencies
• Professional Services
• Retail & Shopping
• Industrial & Manufacturing
• Local Services
• Education & Training
• Government
• Research & Data
• Other

The type describes the site as a whole, not the single page that was cited. reddit.com and youtube.com are Social Media, wikipedia.org is Reference & Encyclopedia, and g2.com is Reviews & Directories, whatever the cited page was about. google.com reads Reviews & Directories because AI answers cite it mostly for local-business listings, while subdomains such as docs.google.com keep their own types.

Known platforms and domain rules assign a site's type first. Sites the rules leave open (those that would otherwise read Other, or rely only on a word in the domain name) are classified in the background from the site's name, its public homepage and a few cited page titles, taking that type only when the classifier is at least 90% confident. Everything else reads Other. Other means we could not place the site with high confidence, not that the site is unusual.

Your own website is never labelled Social Media: if it would be (for example, a company domain with a community or blog-heavy footprint), it reads Other instead.

How Site Type appears across views and exports

• Sources page: the Domains view shows one Site Type column, a Site Type chart above the table, and an All Site Types filter offering the types present in your table.
• Dashboard Over Time: the Source Types chart shows up to 8 slices and folds any remaining types into Smaller types, and each Top Sources row carries the same site-type pill.
• Exports and MCP: the Domains CSV exports Site Type, and the URLs CSV exports both Page Type and Site Type. In MCP, list_top_sources and list_cited_urls return the same 21-type vocabulary in domainType. (One older surface, the blurred By Category preview card on a saved free report, keeps the stored category names from when that snapshot was saved.)

Page Type: what kind of page

In the URLs view, each cited page has a Page Type (with a page-type breakdown chart titled Content Type above the table), such as Listicle, Comparison, Review, Guide, News Article, Blog Post, Forum Thread, Video, Product Page, Home Page, Directory, Documentation or Reference.

A page type can come from two places:

• A classifier that reads the page's address, title and site, where it is available for your workspace.
• A guess from the URL path, used for any page the classifier has not typed or is not confident about. This guess can also show labels such as Category Page, and it falls back to Article.

In a test on 2026-09-25, on 200 production URLs from 192 sites, the classifier matched a blind referee on 86% of pages and the URL-path guess on 34%. Treat a page type as a strong hint, not a fact. See How accurate is AI Peekaboo?.

Owner: yours, a competitor's, or third-party

The Owner column compares each cited site with your brand's website and the websites of the competitors you track:

• You: the site is your brand's website, or a subdomain of it.
• Competitor: the site is a tracked competitor's website, or a subdomain of it. blog.example.com counts for a competitor whose website is example.com.
• Third-party: everything else. It is shown in plain grey text, as the default.

A few rules decide edge cases:

• If a site matches both your website and a competitor's at the same level, it counts as yours.
• A page on a platform, such as a company's Facebook page or link-in-bio page, does not make the platform yours or theirs.
• A competitor tracked without a website can't own anything, so its site shows as Third-party. Add the competitor's website to fix that. See Adding competitors.

What the labels mean, and what they don't

• A type is not a quality score. A Social Media citation is not worth less than a News & Media one. Labels tell you where to act, not how much a citation is worth.
• A type is not an intent. A Reviews & Directories site can host a neutral listing, and a Software & Tools site can host an honest comparison. Open the page before you act on it.
• Owner is only as good as your websites. Ownership follows the websites on your brand and your competitors. A competitor's second domain that isn't its tracked website counts as Third-party.
• Labels don't change your scores. Visibility and citation counts are the same whatever label a source carries.

Using the labels

• Filter Third-party rows in the Domains view with All Site Types to spot places you could earn a citation from: Reviews & Directories, Guides & Comparisons, News & Media and Social Media communities your buyers read.
• Look at Competitor rows to see which of their own sites AI models trust.
• In the URLs view, read the Page Type column to see whether AI models in your category cite listicles, comparisons or guides, then plan content in that format.

See Sources tab for the rest of the page, and Sources and citations for how sources are captured.