New research into how ChatGPT handles comparison questions found a gap that should concern any business counting on AI chatbots to steer customers its way: being mentioned in a source ChatGPT pulls from doesn't mean ChatGPT will actually say your name.
What happened
The analysis looked at how ChatGPT responds to comparison-style prompts — the kind people type when they're deciding between products or services, like "best accounting software for freelancers." Researchers tracked two scenarios: ChatGPT answering from its own training data with no live search, and ChatGPT actively searching the web and citing sources in real time.
In both cases, certain brands were mentioned far less often than their presence in source material would suggest. Even when ChatGPT cited a webpage that explicitly named a brand, the brand itself frequently didn't make it into the actual answer text. The gap wasn't random — brands with less overall visibility in ChatGPT's baseline, no-search answers also tended to underperform when search was switched on, even with equivalent source coverage.
In plain terms: showing up in the footnotes isn't the same as showing up in the sentence. A business can be referenced in an article ChatGPT consults and still get left out of the summary a customer actually reads.
Why it matters
This fits a pattern that's been building since chatbots started citing web sources in 2023. The SEO industry has spent two years scrambling to understand a new scoreboard — not just "do I rank on Google" but "does the AI say my name when someone asks it a question." Tools and vendors focused on this, often grouped under terms like AI visibility or generative engine optimization, have multiplied, but the mechanics of how chatbots decide which brands to surface are still largely a black box controlled by the model makers.
What this data adds is evidence that citation and mention are two separate, loosely connected systems inside these tools. A business's exposure may depend heavily on how prominently it already appears in the AI's training data — essentially, its pre-existing reputation online — rather than on the quality of a single article it's featured in today.
What this means for small businesses
If your business is newer, regional, or simply smaller than the household names in your category, this is worth taking seriously. Getting a mention in a well-ranked blog post or review site no longer guarantees that AI tools will relay your name to a prospective customer, even when that exact page is the source ChatGPT is reading from.
The practical fallout: traditional SEO wins (good rankings, strong backlinks, favorable press) may not translate one-to-one into AI visibility. A business could be doing everything right by old metrics and still be functionally invisible in a ChatGPT answer that recommends three competitors instead.
There's also a cost question. Chasing visibility in AI answers is still an emerging, poorly measured discipline, and most of the tools built to track it are new, unproven, and sometimes expensive subscriptions aimed at larger marketing budgets. Small business owners should weigh the real evidence of return before investing heavily in any single AI-visibility product.
What to watch
Keep an eye on whether OpenAI, Google, or Microsoft publish anything about how their assistants select which brands to name versus merely cite — none have offered meaningful transparency so far. Also watch whether independent brand-visibility trackers start publishing consistent, repeatable methodologies rather than one-off reports, since right now most findings (including this one) come from single studies rather than ongoing benchmarks.
The bottom line
Run your own test this week: ask ChatGPT a comparison question in your industry, with and without browsing enabled, and see whether your business is named — even if you know your content is one of the sources it's pulling from. That five-minute check will tell you more about your real AI visibility than any third-party report.