Metrisque
Blog›You find out by asking the model directly, because…
FINDING16 Sept 2026

You find out by asking the model directly, because standard web signals on your site will not tell you if it recognises your business.

WHAT WE ASKED

how do I know if ChatGPT knows my brand

You find out by asking the model directly, because standard web signals on your site will not tell you if it recognises your business.
The only reliable way to check if ChatGPT knows your brand is to prompt it and see if your name appears in the answer. You cannot deduce it from your website setup or the metadata you feed search engines. In our count across 1,679 products, products with complete schema markup were recommended no more often than products without it. Structured code on a page gave no measurable lift to a brand being named. That creates an uncomfortable blind spot for anyone managing a brand. If you inspect your store pages, your catalog feeds, and your semantic tags, everything might look spotless. The model still might not mention you when a buyer asks for recommendations in your category. The metadata that helps traditional search engines understand your catalog did not translate into higher visibility in our observation. Why this happens remains an open question. Does the training process bypass structured tags entirely, or does the weight of unformatted web mentions simply overwhelm page-level markup? We cannot say from the observation alone. What we can say is that checking your presence requires direct observation. You have to ask the questions your customers would ask, across multiple runs, and count how often your brand appears. There is no shortcut in your site code that tells you the model has logged who you are.

Measured: 1,679 products

Not covered: Memory channel, no browsing. Pre-registered.

Source: https://doi.org/10.5281/zenodo.21417361

Metrisque · LLM Brand Recall · September 2026