We Asked ChatGPT, Perplexity, and Google AI the Same 500 Buyer Questions. Here's Who Got Cited.

We wanted to stop guessing. So over three weeks in July 2026, we ran a controlled experiment: 500 real buyer questions (the kind decision-makers actually type when they're researching a purchase) put to the three answer engines your buyers are using right now. ChatGPT. Perplexity. Google's AI Overviews.

The questions weren't ours. We pulled them from real sales-call transcripts, support tickets, and the "people also ask" data across five B2B and two B2C categories. Things like "best marketing attribution tools for a Series B SaaS," "how do I measure content ROI," and "agencies that do AI search optimisation." Then we logged, for every single answer, which brands got named, and which got ignored.

Here's what we learned. Some of it will change how you think about being found.

Finding 1: The same question gets three different shortlists

The most important takeaway first: there is no single "AI search result." Ask the three engines the same question and you get three different sets of recommended brands, often with barely any overlap. Across our 500 questions, the average brand that appeared in one engine's answer showed up in all three only 19% of the time.

What that means for you: optimising to "rank in AI" is the wrong mental model. You're optimising for three different systems with three different appetites. A brand that dominates Perplexity can be invisible in ChatGPT. If your strategy treats them as one channel, you're leaving two-thirds of the opportunity on the table.

Finding 2: Perplexity cites sources. ChatGPT forms opinions.

Perplexity behaves like a research librarian: it pulls named sources and links them, which means fresh, well-structured, citation-friendly content gets surfaced fast. ChatGPT behaves more like a trusted advisor who has already made up their mind. It draws on a broader, slower-moving sense of which brands "matter" in a category, and it name-drops them conversationally, often without a link at all.

The practical split: to win in Perplexity, you need crawlable, structured, recently-updated content that answers the exact question. To win in ChatGPT, you need to be part of the broader conversation about your category: mentioned, discussed, and referenced across the web enough that the model has absorbed you as a default answer. Different work, different timelines.

Finding 3: The brands getting cited are rarely the biggest

This was the genuinely surprising one. In more than half the categories we tested, the brand cited most often was not the market leader. It was the brand that had published the clearest, most specific, most genuinely useful content on the exact question being asked.

One mid-sized attribution tool got named in 71% of relevant ChatGPT answers (well ahead of competitors with ten times its marketing budget) because it had spent a year publishing precise, honest, deeply technical explainers on measurement. The model learned that this brand explains the topic well, so it became the answer. Money didn't buy that position. Clarity did.

Finding 4: Most brands have never checked, and it shows

For the brands in our test that weren't getting cited, the causes were almost always fixable: content that talked about themselves instead of answering the buyer's question, pages that AI crawlers were quietly blocked from reading, and a total absence of the structured, declarative, "here is the answer" writing that engines reward. None of it was exotic. It was just work nobody had done, because nobody had looked.

What to actually do with this

Start by looking. Take your ten highest-intent buyer questions and run each through all three engines. Write down who gets named. If it's not you (and for most brands, in most categories, it currently isn't), that gap is the single clearest picture you'll get of where your next customers are being sent instead.

Then decide whether being the cited answer is worth building toward. For a growing number of categories, the brands that win the next five years won't be the ones that rank first on Google. They'll be the ones the AI recommends by name, before the buyer ever opens a search tab.

That's the work we do. If you want to see your own category's results, the audit is where we start.