Brands are paying for bad product data twice now

Adobe just reported that 34% of retail product pages can't be properly read by AI, in their Q1 report.

This is despite the fact that AI-driven traffic has been growing for over a year, and the same report also confirmed that the AI-driven channel converts 42% better than traditional channels.

Most CMOs have seen numbers like this by now, but you don't really see budgets shifting in response. I think part of the reason is simple: until recently, the cost of bad product data was basically invisible.

For years, you could get away with thin product pages. Missing attributes, inconsistent naming, vague descriptions, it didn't matter that much. You'd still rank in search, still show up in retail media, still convert at roughly normal rates. The distribution layer did a lot of the heavy lifting.

So nothing in your reporting ever told you there was a problem. Traffic came in, revenue came in, and that was enough. But the gap was always there, it just showed up in ways you couldn't easily measure, like weaker matching or less relevant traffic.

What's changed with AI-driven answer engines is that they rely directly on the product data. When it is vague, inconsistent, or incomplete, it becomes harder for them to match products to shopper intent. And if that match doesn't happen, your product just never gets considered in the first place.

So a brand now pays for its product data twice: once through the conversion on the traffic that got through, and once through the recommendation it lost on the traffic that didn't.

The brands acting now aren't the ones who suddenly believe in AI. They're just realizing their product data hasn't been as strong as they thought, and this is the first time a channel has made that hard to ignore.