Citations

Most people walk into the Citations page expecting Google. They think "featured snippet", they think "backlinks", they assume more is better and that the page they're looking at is roughly an SEO report with a coat of paint.

It isn't. Citations is a different sport, and being cited often matters more than being ranked. Before you click around the page, it's worth getting that distinction right.

What "being cited" actually means

A citation is a URL that an AI model returns as a source for its answer.

When you ask Perplexity "best CRM for small business," it composes a short answer and then lists a handful of links underneath. Those links are the citations. They're the receipts: the pages the model went and read, the sources it leaned on, the places it's quietly telling you "this is where I got this."

ChatGPT does the same thing when it's in web-search mode. Google does it under every AI Overview, with a little stack of reference pages attached. Different surface, same idea.

Trakkr collects every citation every AI model returns for every prompt you track, deduplicates them, classifies them, and stitches them into one searchable view. That view is what the rest of this page is about.

A few things citations are not:

• They're not just brand mentions. A page can cite you (the model used it as a source) without naming you in the answer, and a model can name you in the answer without citing anything at all (it's working from training data).
• They're not Google rank. The position on a citation list is not a SERP position. Models pull sources in roughly the order they used them, but small shifts in phrasing reshuffle the list. The list itself is the signal.
• They're not links to you. Most of your citations will be pages on other sites that happen to talk about you. Your own domain is one source among many, and usually not the biggest one.

Why citations are a different game than SEO

The unit is different. The mechanic is different. The decision they inform is different.

Here's the implication that catches people off guard. In AI search, the page that ranks for the query is often not the page that gets cited. A model asked "best running shoes for marathons" doesn't open the top SERP and copy it. It pulls a Wirecutter review, a Runner's World guide, a Reddit thread, and a couple of brand pages, and writes an answer from the lot. Your SEO rankings barely enter the picture.

So your visibility in AI is a downstream effect of your citation profile: which sources mention you, how prominently, in what context. Sort that out and AI visibility follows. Ignore it and you can rank #1 on Google and still be invisible in ChatGPT.

Where Trakkr gets citation data

Three AI surfaces return sources alongside their answers. Trakkr captures all of them.

Each of these platforms runs every active prompt against your category once a day. Every URL that comes back goes through the same pipeline: we strip tracking parameters, normalize the domain, dedupe across runs, fetch the page, classify the source type, score sentiment toward your brand, and pull out any competitors named on it.

The other models you see elsewhere in Trakkr (Claude, stock ChatGPT without web search, Gemini in pure-LLM mode) don't return citations. They influence visibility, but they don't show their work, so they don't appear here.

How citations get classified

The same URL means different things depending on what it is. A G2 review and a TechCrunch article both count as "a citation," but they should be read very differently. Trakkr tags every cited domain into one of eight source types:

The source type matters because models weight them differently and so should you. Earned media and institutional citations carry the most lift, reviews are the workhorse for product-led categories, social signals show what real users say but can swing wildly, PR wire is the lowest-leverage type and often noise.

Citation rows also carry sentiment (positive, neutral, negative toward your brand), the prompts that produced the citation, the competitors that show up on the same page, and the dates the URL was first and last seen.

Reading the Citations page

The page has five working views. They all share the same brand, same filters, same tag scope. Each view points the same data at a different question.

The four to know are Sources, Pages, Feed, and Queries. Heatmap exists as a legacy view if you used it under an earlier version, and Outreach gets its own page.

Sources: who's talking about you

The default view. Every citation gets grouped by domain, so you're reading the publications and websites, not raw URLs. Click any domain to open its Source Profile in the right pane: citation count over time, the specific pages on that domain that are being cited, which prompts triggered them, the sentiment toward your brand, and where you sit relative to competitors on this same source.

The most useful filter here is the Gaps tab. It surfaces domains where competitors are cited and you aren't, which is the cleanest definition of "where you should be." Stack it with the Earned media or Review source type and you have a tight list of high-leverage targets.

This is the view to live in when you want to answer which publications matter for my category, and how am I doing on each one?

Pages: the actual content driving the answer

Sources groups by domain, Pages flattens to individual URLs. The same domain might have one canonical guide that's responsible for half the citations and a dozen unrelated pages that contribute almost nothing. Pages shows you that.

Filter to Your Pages to see which pieces of your own content AI is actually using as source material. That list is short for almost everyone, and it's a useful corrective. The page you spent three weeks writing might not be on it. The throwaway comparison post from 2023 might be your single best AI asset.

Feed: what changed

A chronological changelog of citation events. Four event types:

Read the Feed weekly, on the 7-day window. It's the fastest way to catch a sudden drop in coverage, a competitor moving in on a page you used to own, or new wins from a recent PR push.

Heatmap: the legacy grid view

A matrix with citation sources down one axis and brands across the other, with cells colored by who's cited where. Useful for executive screenshots and for spotting category-wide patterns at a glance ("I lose every review site, win every news site").

It's hidden from the main tab bar now (most of what it showed lives more cleanly in Sources with the gaps filter), but it's still reachable at ?view=heatmap if you bookmarked it.

The query angle: which prompts produced each citation

Every citation in Trakkr is traceable back to the prompts that triggered it. Open the Queries view and you flip the perspective: instead of "what's citing me," you're reading "what questions cause these citations to surface."

Trakkr classifies each query by intent:

This matters for content strategy because each intent is satisfied by different kinds of pages. Discovery queries reward roundups and "best of" articles. Comparison queries reward head-to-head pieces. Alternative queries reward "X alternatives" content. If you're missing from Alternative queries, the gap usually isn't a brand problem, it's a missing piece of content somewhere.

The two filter tabs to know:

• Citing shows the queries where you're already in the citation list. Read these to understand what's working and where you'd defend.
• Gaps shows queries where competitors are cited and you aren't. Sorted by opportunity score, these are usually your shortest path to new coverage.

From any gap row, you can promote the query straight into your tracked Prompts list (so you watch it daily going forward) or hand it to the content composer, which pre-fills with the query, intent, and the competitors winning it.

Working the data

The Citations page is a diagnostic, not a deliverable. The work is what you do with what it tells you.

A reasonable weekly cadence for most teams:

1. Open Sources, filter to Gaps + Earned media + Review. Read the top three domains. For each, click in and look at the specific pages that are cited. That's where to focus outreach or content investment.
2. Open Queries, filter to Gaps, sort by opportunity score. Pick one or two and either add them to your tracked prompts, or build a piece of content that targets them directly.
3. Open Feed, 7-day window. Scan for lost coverage and new competitors. If anything looks bad, click through to the source and read it.

If you want a working queue of high-leverage targets ranked by fit and difficulty, head to Outreach. It takes the gaps Trakkr surfaces and turns them into a prioritized list with pitch angles and status tracking.

Common questions

What's the difference between a citation and a mention?

A citation is a URL the model returned as a source. A mention is your brand name appearing in the answer text. They overlap but they're not the same. A glowing Wirecutter review can be cited (the model used it) without your brand being named in the synthesized answer; a model can name you from training data without citing anything. The Citations page tracks the URLs. For answer-text mentions, that's Perception.

Why is one of my pages cited a lot and another isn't?

AI models pick sources that look authoritative and on-topic for the specific query. Long-form, structured, well-linked content tends to win. Short pages, marketing pages, and pages without clear answers tend to lose. The Pages view filtered to Your Pages will show you which of your own work is doing the heavy lifting.

My competitors are cited on a domain I've never heard of. Should I care?

Sometimes. Filter that source's profile and look at how many citations it's producing and how much authority it carries. A niche blog cited five times is noise. A vertical publication you didn't realize existed, cited fifty times across your category, is a real lever. Source type and citation count together tell you which it is.

How fresh is the data?

Citations refresh once a day, aligned with the daily prompt runs. The Feed view always shows changes against the previous snapshot, so the difference between today and yesterday is one click. If you've just done a PR push, expect the new citations to appear within a day or two, faster on Perplexity (which retrieves live), slower on ChatGPT Search and Google.

Why is Perplexity showing more citations than the others?

By design. Perplexity returns 5-10 citations for almost every answer; Google's AI Overviews cite a handful of references; ChatGPT Search varies. So Perplexity tends to dominate raw citation counts. Read the Provider filter as a coverage signal, not a popularity contest, the question is whether each surface is citing you at all, not which has the bigger number.

Can I get cited just by writing more content?

Indirectly. Models cite the sources they trust on a topic. If those sources are reviews, no amount of your own blogging changes that, you need to be on the reviews. If those sources are roundups and guides, your own well-structured content can get cited directly. The Sources view tells you which kind of category you're in before you decide how to invest.

Where do I act on citation gaps?

Two places. The Outreach view turns gaps into a working queue with pitch angles. The content composer (under "Create content" on any gap query) drafts pages targeted at the queries where you're losing. Both feed back into citation tracking, so you watch the gap close in the Feed once the work lands.