Visibility in AI without the conversation that came before

Translated from the original Spanish article.

Some companies want to be part of the conversation about how to gain visibility in AI without ever having wanted to take part in the conversation that came before it.

Last week Webflow published a study in which they analyzed two thousand company websites in the United States and measured how ChatGPT, Gemini, Perplexity and Claude mention them.

The data is interesting: the typical brand shows up in only 16% of the answers relevant to its category, that is, across the set of questions where that company should be a legitimate candidate. And it gets a citation link 6% of the time.

It’s true that those two figures call for some caution, because the report is signed by the people selling the solution, and it doesn’t publish the questions it put to the models.

In any case, what interests me is the relationship between the results and the quality of the company website. Because, according to the study, when AI mentions a brand, it describes it incorrectly about a third of the time.

What you find when you open the hood is that 62% of those websites have broken internal links, 60% lack basic metadata and 54% haven’t refreshed their content in the last six months.

None of that belongs to the new paradigm. And, at bottom, it’s a question of priorities.

Another study, by G2, points in the same direction: it reports that 51% of B2B Software buyers start their research with AI, which is now the main source of influence on their decision about which vendors make the shortlist.

How much is it worth to a brand for the models to recommend it for the relevant queries in its category? And for them to offer accurate, complete information about it?