Skip to main content

Understanding citations and sources

The pages shaping what AI says about your category — how they're grouped, classified, and scored.

A citation is a web page an AI assistant referenced in its answer. Citations are the most actionable data in MentionBird, because they show why an answer came out the way it did.

If a Runner's World roundup is cited in most marathon-shoes answers and Nike isn't in that roundup, you've found something you can act on. Your visibility number tells you that you're losing; the citations tell you where.

The domain leaderboard

Citations are grouped by domain, ranked by how often each is cited. This is your map of the sources that carry weight in your category.

Expect three broad kinds: your own pages, competitors' pages, and third-party sites belonging to neither. That last group is usually the largest, and usually where the leverage is. See Owned, competitor and earned sources.

Click any domain to see the specific URLs cited from it, which matters because "runnersworld.com is cited a lot" is much less useful than knowing exactly which article. Drilling into a single domain covers how to read that view.

Two classifications

Every cited page carries two labels answering different questions.

Source type — what kind of site is this? Applied per domain, one label each: news and media, forums and communities, review and comparison sites, marketplaces and aggregators, blogs and publications, encyclopedias and reference, documentation, help and support, corporate sites, government and institutional, academic and research, social.

Subdomains are classified on their own merits, so a company's documentation subdomain is labelled as documentation rather than inheriting the corporate label from its parent.

Content type — what kind of page is this? Applied per URL, and a page can carry more than one: listicle, comparison, how-to, review, case study, news, research, plus structural types like product, documentation, community, reference and landing.

The distinction earns its keep when you filter. "Forums" tells you Reddit matters in your category. "Listicles" tells you the "best running shoes for X" format matters — a different insight, and a different piece of work.

When you see "misc"

A misc label means the classification couldn't be determined confidently for that page — it's an honest unknown rather than a category. Reading it as "miscellaneous sites are important in my category" would be a mistake.

Domain authority

Each domain carries an authority score from 0 to 100, where higher means a more widely-cited site. It's useful for triage: a citation from a high-authority publication is generally a bigger win than one from an obscure blog, and a high-authority page you're missing from is usually worth more effort.

A blank score means unknown, not low. Newly-seen domains are unscored until they've been looked up, and some never resolve. A blank is shown as a dash precisely so it isn't misread as zero — never conclude a site is unimportant because its score is empty.

Searching and exporting

You can search citations by domain, URL, or page title. Two common uses: checking whether a specific publication cites you at all, and finding every citation of your own site to see which of your pages assistants actually reach for.

Whatever the page is showing — filtered, searched, or scoped to one prompt — can also be downloaded as a CSV. Exporting citations and sources covers the formats and what each is for.

How to use this

The workflow that produces results:

  1. Find a prompt where you're losing.

  2. Look at what's cited in those answers.

  3. Check whether you appear on those pages.

You'll usually land on one of two conclusions: a page of yours is being cited but isn't persuading, or a third-party page is deciding the answer and you're absent from it. Those need different responses, and the citations are what tell you which one you have.

Did this answer your question?