I tested one buyer question across ChatGPT, Perplexity, and Google. The results surprised me.
I'm researching how often AI answer engines recommend competitors instead of a company.
As a test, I used this buyer question:
"What is the best customer feedback tool for a small B2B SaaS company that needs account-level revenue context?"
ChatGPT, Perplexity, and Google all recommended competitors.
One actual company I checked never appeared in any of the results.
That got me wondering:
How many SaaS companies are missing from buyer-facing AI recommendations without realizing it?
For founders here:
Have you ever checked whether ChatGPT, Perplexity, or Google mention your company for the buyer questions that matter most?
If you did check, what would you actually want to know?
Run the same five buyer questions in fresh chats across each engine, then log the date, question, named companies, and cited sources separately. The useful signal is the pattern over several weeks, not one mention. If you have not checked yet, a free five prompt snapshot is a simple place to start.
For that query, I'd be interested in which tools actually support account-level revenue context and what sources back that up. That would help me understand how clearly each product communicates this capability.
Did exactly this two days ago, six engines and five questions, on our own domain.
The surprise was not absence. It was that presence told me nothing. GPT named us in five answers out of five, which reads as a win until you look at what it cited, which was almost entirely our own pages and not one third party article. Perplexity meanwhile refused us on the list question because our site was not in its sources.
So to your second question, what I would want to know is not whether I am mentioned, it is what got cited. That is the part that gives you the mechanism. Only your own pages means you need third party material. Only directories means you need editorial rather than more listings. Nothing at all means the model is asserting you without evidence, which is the most fragile of the three.
Two cautions from doing it. Keep branded and unbranded questions apart, they have different mechanisms and different fixes, and yours is correctly the unbranded one. And run each question several times, because a single pass is not deterministic and a company that surfaces a third of the time will look absent on one run.
That's a good point. The source being cited may be more important than simple presence/absence. One thing that surprised me was how different the results were across engines for the same query. I'm starting to think the most useful part of the audit might be identifying which sources are getting cited and why, rather than producing a visibility score.