When We Asked ChatGPT About Ourselves, It Recommended Someone Else

We built Citovue to measure whether AI assistants cite and recommend a company when someone asks a buying question. Before we sold that to anyone else, we ran our own methodology on ourselves. What we found on ChatGPT wasn't obscurity. It was misdirection.

What we did

On 2026-07-21 we ran our own 20-prompt AI-visibility audit against Citovue, on ChatGPT, in a single logged-in session with no memory or custom instructions turned on — the setup our methodology uses to approximate how a real prospect would ask. Five of those twenty prompts (our “Category B”) ask ChatGPT to directly compare “Citovue” against a named competitor, Citable, worded the way a real buyer actually would: “Citovue vs Citable — which one should I use for an AI visibility audit?” and four close variants.

What we found

Citovue was never cited — 0 of 20 prompts, across every category we tested. That wasn't the surprising part. The surprising part: in four of the five direct comparison prompts, ChatGPT didn't just fail to find us. It silently substituted a different, real, unrelated company with a similar name — “CiteVue” (citevue.com) — and answered the whole question as if we'd asked about them.

The first prompt makes the swap explicit before anything else:

From there it built a full comparison table and recommended “CiteVue” for exactly the scenario that is Citovue's actual product — “a client audit deliverable quickly,” “a baseline score and a list of fixes” — without ever noting that a company named Citovue exists. One prompt later, asked which is better value, ChatGPT quoted a specific price for “CiteVue”: $19.99. Our real audit price is £1,500. Not a rounding difference — a different product, a different currency, a different order of magnitude, attached to our name.

The fifth prompt pulled in a third real company, “Citovo,” and ChatGPT flagged its own confusion directly: “There is some name confusion here.” Even so, it never considered that “Citovue” might refer to a company distinct from the three it had found.

Only one prompt avoided this. It was the one framed around “retainer” rather than “audit” — there, ChatGPT said plainly it couldn't verify a real Citovue and reasoned under our actual name instead of borrowing someone else's. That's a useful clue: the substitution looks tied to how the question is phrased, not just the name alone. We want to be precise about what this is and isn't: one platform, one day, one 20-prompt run. We haven't tested whether this holds on other AI assistants or on a different day, and we're not claiming it does. But in this run, it happened in 4 of 5 direct comparison prompts, and once with a specific, fabricated, wrong price attached.

To be clear: CiteVue and Citovo are real, legitimate companies that did nothing wrong here. This isn't about them. It's about what ChatGPT does when two names sit close enough to blur.

Why it matters

Most writing about AI visibility talks about being invisible — absent from the answer, a blank space where you should be. That's real, and it's the problem we built this company to measure. This finding is a different, sharper problem: a prospect who already knows your name and types it into ChatGPT with clear buying intent can be actively redirected to someone else's product, at someone else's price, with neither company doing anything wrong. Being unranked costs you an opportunity you never had. Being misidentified hands that same opportunity to a name that merely sounds like yours.

If your brand shares a naming pattern with competitors — a common prefix, a common suffix, something close enough to read as a plausible typo — this is worth checking, not assuming away.

What we're doing about it

This finding is now a formal part of our own audit methodology, not a work in progress. “Misidentified as another brand” is a documented citation category — with its own severity notes, versioned as v1.2 — because it's a worse outcome for a business than simple absence. It earns its own line in every report we run, not a footnote under “not cited.” On our own site, we're building direct disambiguation content and clearer entity signals so AI systems have less room to guess wrong about who Citovue is.

We've also finished running this same self-audit across our full five-platform panel — ChatGPT, Perplexity, Microsoft Copilot, Claude.ai, and Google AI Mode. Short version: the CiteVue substitution wasn't unique to ChatGPT. Perplexity, Claude.ai, and Google AI Mode each made the same swap independently, in their own separate sessions — and three of those four platforms (ChatGPT, Perplexity, and Claude.ai) landed on the exact same fabricated price, $19.99. Each platform fails in its own distinct way beyond that; how each one specifically fails is a separate piece, not this one.

If you want to know whether something similar is happening to your brand, email [email protected] — that's exactly what an audit is built to catch.