tl;dr: To check if AI tools are citing your content, run the same prompts your buyers would use in ChatGPT and Perplexity. Then look for two things: whether your brand is named in the answer, and whether your URL shows up in the source list. Mentions don’t drive traffic. Citations do.

Most marketing teams checking AI search are still checking the wrong thing.

They run a branded query like “tell me about [their company],” see the AI describe what they do, and feel reassured. Or they ask “what are the best [their category] platforms?” and feel relief when their brand name appears in the answer. Then they wonder why traffic from AI isn’t showing up in analytics.

Here’s what I think: a mention is good for awareness. A citation with a clickable source link is what drives qualified visitors to your site. Most teams are still optimizing for the first when they should be measuring the second.

The good news: you can see exactly where you stand in about 10 minutes. Here’s how.

What you need before you start

You need three things to start:

• Access to ChatGPT and Perplexity (free versions work fine)
• A list of 3 to 5 prompts your buyers would realistically type into an AI tool
• A notes doc or spreadsheet to track what you find

A citation, for the purposes of this exercise, is when an AI tool links to your URL as a source for its answer. A mention is when the AI names your brand inside the answer text without linking. The exercise is designed to surface both, then separate them.

Step 1: Write down the prompts your buyers would actually use

Skip the branded ones. “What does [your company] do?” tells you nothing useful. Your buyer already knows your name before they type that.

The prompts that matter are the ones a prospect types when they’re researching a problem you solve. Some examples for an insurtech audience:

• “What are the best claims automation tools for mid-market carriers?”
• “How do MGAs handle compliance for embedded insurance products?”
• “Compare [category] platforms for commercial lines brokers”

Aim for 3 to 5 prompts that reflect commercial intent (someone evaluating solutions) rather than informational intent (someone learning a topic for the first time). Commercial-intent prompts are where citations actually convert.

Step 2: Run each prompt in ChatGPT and Perplexity

Open both tools. Use a fresh chat for each prompt so prior conversation doesn’t influence the answer.

For each prompt, capture:

• Whether your brand is mentioned in the answer text
• Whether your URL appears in the source list or citations panel
• Which competitors are mentioned
• Which competitors are cited with linked sources

Perplexity makes citations easy to spot because they’re listed numerically inline and in a sources panel. ChatGPT shows sources differently depending on whether web search is enabled. Make sure web search is on, otherwise you’re querying training data, not the live web, and the answer won’t reflect what your buyers actually see.

Step 3: Sort the results into mentions versus citations

This is the step most teams skip, and it’s the whole point.

Build two columns: “mentioned” and “cited.” Put each occurrence of your brand into one column or the other.

• Mentions matter for awareness. They tell you the AI knows you exist as part of the category conversation.
• Citations matter for traffic, because they’re the only result that gives a reader a clickable path to your site.

If you have 5 mentions and 0 citations, you have a visibility problem disguised as a win. Your brand is being talked about by AI tools without sending you any visitors.

Step 4: Look at who is getting cited and figure out why

This is where the exercise stops being a check-in and starts being a strategy input.

For every citation that didn’t go to you, click through. Read what the cited page actually says. You’re looking for patterns:

• Is the answer to the prompt in the first paragraph?
• Are claims attributed to named sources or named methodologies?
• Are headings shaped as questions a buyer would actually ask?
• Is the page structured so individual sentences could be lifted out as standalone answers?

Most pages that get cited share those traits. Most pages that don’t, share the opposite ones: throat-clearing introductions, generic claims with no attribution, headings written for SEO instead of for how people prompt AI tools.

What this exercise actually tells you

If you’re getting cited regularly, your content is doing the structural work AI tools reward. That’s a real signal, not luck. Track which pages drive the citations, then look for the pattern you can replicate across the rest of your library.

If you’re getting mentioned but not cited, you’ve already done the hardest part: built enough brand recognition that AI tools include you in the category conversation. The citation gap is a structural fix, not a brand-building project. Most of the lift comes from how you write, not how loud you market.

If you’re getting neither, you’re early. That’s not a problem, that’s a window. Most insurtech teams are in the same position right now, which means the first ones to structure their content for citation are going to set the standard in their category.

What to do with what you find

Run this exercise once a quarter, not one-and-done. AI tools change how they pull citations frequently, and competitors are adjusting their content, too.

Here’s the part most teams miss: the exercise isn’t the strategy. It’s the diagnostic. Once you know which bucket you’re in, the next move is clear. Either you have a structure problem you can fix, or you have a visibility window to step into.

The question worth sitting with isn’t “are we showing up in AI answers?” It’s “when buyers ask the AI a question we should be answering, whose page does it pull from, and what is that page doing that ours isn’t?”

Frequently asked questions