How to Find the Sources AI Cites About Your Brand, via API

When ChatGPT recommends a tool in your category, it did not make that up. It read some pages and repeated what they said. Those pages are the citations behind the answer, and if you want to change what AI says about your brand, those pages are the thing you actually have to change.

Most people obsess over the answer itself. Whether they got mentioned, whether a competitor did. That is worth knowing, but it is downstream. The real lever is the set of sources the model keeps pulling from. Get cited on those, and the answer starts to move on its own.

The problem is you cannot see them. You can ask ChatGPT the same question fifty times and take notes, but you will never get a clean, ranked list of which domains and pages are doing the work. That is what the Trackee Citations endpoint gives you in one call.

What citations actually are

A citation is a page an AI model links to as a source when it answers a question that mentions your brand. Ask "best CRM software" and the model leans on a handful of pages to build its answer. Some are review sites, some are Reddit threads, some are the vendors' own docs.

Citations aggregates all of those across every answer it can see, then ranks them. You find out which sources shape the conversation about your category, and how much of that conversation each one owns.

If you have never looked at this before, run your brand through the free AI Citation Checker first. No code, no key, just type a brand and see the sources in your browser. Once it clicks, come back and wire up the API.

What you need

Three things, same as any Trackee endpoint:

• A free Trackee account. The free plan covers testing.
• An access key from your dashboard.
• The x-access-key header on every request.

No SDK, no OAuth, nothing to install.

Step 1: Get your API key

Sign in to your dashboard and open the API keys section. Create a key and copy it right away, since the full value only shows once.

That key goes in the x-access-key header on every call. Keep it server-side. Treat it like any other secret.

Step 2: Ask for the citations

Here is a full request. It asks for the domains AI cites most when it talks about Salesforce. Swap in your own key.

curl -X POST https://api.trackee.dev/v1/citations \
  -H "x-access-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "brand": "Salesforce",
    "group_by": "domains"
  }'

The response comes back like this:

{
  "success": true,
  "credits": { "charged": 35, "remaining": 4930 },
  "data": {
    "brand": "Salesforce",
    "group_by": "domains",
    "total_mentions": 551038,
    "citations": [
      { "key": "www.salesforce.com", "mentions": 19169, "share": 0.035 },
      { "key": "www.reddit.com", "mentions": 15877, "share": 0.029 },
      { "key": "www.g2.com", "mentions": 9250, "share": 0.017 }
    ]
  }
}

Each query costs 35 credits, and you can see that in credits.charged. credits.remaining is your balance right after. So you always know exactly what a call spent.

Reading the response

Everything useful lives in data.

brand and group_by are echoed back, which helps when you are firing off a batch of calls and logging what came back.

total_mentions is the total citation volume behind the whole picture. In the example, 551,038 citations across every answer that touched Salesforce. That number gives the shares below something to divide into.

citations is the ranked list. Each entry has three fields:

• key is the source. Here it is a domain, because we grouped by domains. Grouped by pages it would be a full URL.
• mentions is how many times that source got cited.
• share is that source's slice of all citations, as a decimal. 0.035 means this domain is behind 3.5% of everything AI says about the brand.

Look at that example again. Salesforce's own site is the top source, which you would hope. But Reddit is right behind it, and G2 is close after. That tells you a lot. A big chunk of what AI repeats about Salesforce comes from a forum thread and a review site, NOT from Salesforce. You do not control those pages, but you can influence them.

Pages or domains

The group_by field flips the whole view.

Group by domains to see which sites shape your category. Reddit, G2, a couple of big publications. This is the map of where the conversation happens.

Group by pages to get exact URLs instead. Now you can see the single review post or the one comparison article that keeps getting cited. That is a concrete target. You can pitch an update, correct a fact, or earn a spot on that exact page.

curl -X POST https://api.trackee.dev/v1/citations \
  -H "x-access-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "brand": "Salesforce",
    "group_by": "pages"
  }'

Start with domains to understand the landscape, then drill into pages once you know which sites matter.

Why this is the GEO lever

GEO, generative engine optimization, is the same idea as SEO but the target is AI answers instead of blue links. And citations are how you pull that lever.

Here is the loop. Run Citations on your own brand and on your top competitor. Compare the lists. The sources cited heavily about them but not about you are your gap. That is where you go earn a mention, a review, a mention in a roundup, a helpful Reddit answer. Do that, and the next time the model answers, it has a reason to include you.

This is different from checking whether you got mentioned at all. Mentions tell you the score. Citations tell you why the score is what it is, and what to change. If you want the mentions side too, I wrote a separate walkthrough on tracking brand mentions via API that pairs well with this one.

No code? Same data in the dashboard

If you would rather not touch curl, all of this runs in the dashboard with no code. Agencies use it to watch citations across a whole roster of brands in one place, which is hard to do by hand across that many clients.

And if you live in an AI editor, Trackee ships an MCP server and a coding skill so an agent can pull citations in plain language. You ask "which domains does AI cite most about Salesforce," and it makes the call for you.

What's next

You can now pull the exact sources AI leans on for any brand, ranked by mentions and share, grouped by page or domain.

A few good next steps:

• Read the full AI Citations API page for every parameter and field.
• Browse the API reference for the rest of the endpoints.
• Start on the free plan and scale up when your query volume grows.

Run your brand and your closest competitor, put the two lists side by side, and find the three sources cited about them but not you. Those three are your GEO to-do list.