I was shopping for a cat fountain and accidentally found an AI visibility pattern

I’m learning AEO/GEO by running small experiments in public instead of just reading opinions about AI search.

This one started in the least glamorous way possible: I wanted to buy a cat water fountain.

I opened a few review pages, asked Perplexity and ChatGPT, and quickly realized I wasn’t only comparing products. I was comparing the sources AI used to decide what a brand “is.”

That made me curious:

When someone asks AI about a specific brand, does the brand’s own website control the answer?

I tested 15 queries around one product category, using PETLIBRO as a public example. The queries were split into three layers - here’s what stood out:

Problem-aware queries
Example: “why won’t my cat drink from a bowl?”
Result: the brand disappeared. AI cited vets, health sites, Reddit, and pet-care blogs.

Solution-aware queries
Example: “best / quietest cat water fountain”
Result: the brand appeared often as a top pick. AI cited review media, retailers, and brand pages.

Brand-aware queries
Example: “[brand] review / vs competitor / alternatives”
Result: the brand appeared, but the answer was shaped by third-party sources too.

The interesting part was the branded layer.

Even when the query was explicitly about the brand, AI didn’t only use the brand’s site. It pulled from review sites, retailer reviews, Reddit, YouTube, and app/review signals.

My takeaway:

A brand’s website makes claims. Third-party sources make those claims believable. AI seems to use both.

That changed how I think about AEO.

Before this, I would have mostly checked:

• are the pages crawlable?
• are product pages clear?
• is there schema?
• are comparison pages written?
• are the core claims stated clearly?

I still think all of that matters.

But now I’d also audit:

• which review sites AI repeatedly cites
• whether Reddit appears
• whether retailer reviews shape the answer
• whether YouTube tests exist
• which caveats AI repeats
• which query layer the brand disappears from
• which attributes AI gives to competitors instead

This is the part I think founders should care about:

AI visibility is not just a website/content problem. Sometimes the fix is product, support, reviews, positioning, community, or third-party proof.

Small caveat: this was a tiny test, not a benchmark. 15 queries, two engines, one run each, visible citations only.

I cleaned the raw outputs into a source-level dataset instead of publishing the full AI responses. The cleaned version keeps query, engine, brand role, cited domains, source type, gap, and notes.

Full write-up + cleaned 30-row dataset:
https://www.nixal.ai/blog/ai-brand-visibility-brand-queries-source-gap/

I’m curious if other founders are checking this yet:

When people ask ChatGPT or Perplexity about your product/category, does AI mostly cite your own site, or does it lean on third-party sources for trust?