I’m beginning to believe schema influences ChatGPT citations. For this eComm prompt set, MANY of the most cited sites use Product schema.

At Nectiv, I’ve been having a lot of fun playing around with the AI Tracker that we’ve built. We’re tracking a few different larger prompt sets for different industries including software and ecommerce queries. It’s been cool to analyze the types of sites that get cited and different trends in ChatGPT.

One thing that really stood out to me when analyzing our eCommerce prompt set. In the AI Tracker, we’re analyzing 500+ prompts that are pulled from Amazon data. The goal is to give us a high level view of what influences results in the commerce space.

When analyzing this prompt set, I was surprised at some of the most commonly sites domains. Many of these were editorial affiliate sites that were topically unrelated to commerce. Some of these included sites like NY Mag (3.0%), People (2.3%), FindThisBest (1.9%), Esquire (0.5%).

When analyzing those site, one common thread is that the vast majority markup their editorial articles are marked up with Product structured data. Many of them use Product schema directly, while others embed it as a child element under a parent schema (NewsArticle). But it’s there more oftentimes than not.

FindThisBest is particular interesting. This is a site that has been completely banished from Google. Since the Helpful Content Updates have started rolling out, this site has gone to basically zero in Google search. However, it finds itself in the top 10 results for this set of commerce queries.

Obviously, there could be many different factors going on, but it’s very interesting to me that the editorial sites using Product schema, tend to get included in ChatGPT citations. This is making me thing that structured data elements might have a bigger role in ChatGPT citations after all.