The AI engines name my competitor in 99% of answers. Me: 17%.

Some of my customers found Prometora through Google. But the newer ones increasingly tell me a different story: they asked ChatGPT "what should I build my marketplace on?" and went down whatever list it gave them. I have written before about the crawler side of this, the bots reading my pages at 3am. This issue is about the other side: when a buyer asks the question, whose name comes back?

Until this week I only had a feeling. Now I have a number, and the number is humbling.

I wrote down the twelve buying questions I believe my customers ask an AI. Not keywords, actual questions: "What are the best alternatives to Sharetribe for building a two-sided marketplace?", "What is the best platform to build a rental marketplace, like an Airbnb for equipment or spaces?", "Which marketplace software handles seller onboarding and automatic payouts to sellers via Stripe Connect out of the box?" and nine more like them.

Then I built a small script that asks each question three times per engine, through the official APIs with web search switched on, and records two separate things for every brand I track:

Two honesty notes before the numbers. First, this is a poll, not a census. Nobody can see real user conversations, me included. Asking the same questions on a schedule and counting is the best available proxy, and answers vary between runs, which is exactly why each question is asked three times. Second, the API with web search is a stand-in for the consumer apps, not a perfect mirror of them.

Six brands tracked, twelve questions, three runs each, two engines that completed: 72 answers.

Sharetribe was named in 71 of 72 answers. It missed exactly once.

One caveat, Claude's web-search calls failed (a tool error on my side of the fence, since fixed), so its 36 answers came purely from what the model remembers, with no sources attached. The ChatGPT column did search properly. I am publishing the day-one numbers as they happened rather than quietly re-running until they look cleaner.

On the direct question, "Prometora vs Sharetribe - how do they compare?", both engines named Prometora in 3 out of 3 runs, and ChatGPT cited my site every time. The engines know who I am, and describe the product accurately when asked.

On the open questions, where the buyer has not named me first, I mostly vanish. "Best platform for a rental marketplace": 0 for 6. "Marketplace where customers book services from local providers": 1 for 6. "Marketplace software with Stripe Connect payouts out of the box": 0 for 6, which stung the most, because payouts via Stripe Connect is the topic I have written and filmed more about than any other.

So the gap is not awareness. The engines have read my site. The gap is between being in the index and being volunteered as an answer.

This is the finding I did not expect. ChatGPT named Prometora in 8 answers, and in all 8 it cited a source for that mention. Every single citation was prometora.com. Not one third-party page, in 36 searched answers, caused an engine to bring me up.

Then look at what the engines cite overall on these questions: the vendors' own domains first (sharetribe.com 29 times, dokan.co 16, cs-cart.com 14), and then a neutral layer of sites they appear to treat as trustworthy context: stripe.com, shopify.com, wordpress.org, bubble.io. The comparison articles, directories and community threads that would recommend a newer product barely show up for me at all, because for my brand they barely exist.

Your own website gets you cited when you are already named. It is other people's websites that get you named.

The script now runs every Monday morning and commits its numbers, so this becomes a time series whether it flatters me or not.

Around mid-September 2026 I will publish the follow-up: same twelve questions, same method, and we will see whether the 17% moved. If it did not, you will read that too.

- Rasmus