How to prepare product feeds for AI shopping agents without ignoring product pages

Better product feeds are worth the work. Neglecting the product page because an AI agent supposedly never opens it is not. That is where DataFeedWatch pushes a useful argument too far. OpenAI explicitly says its shopping research reads product pages as well as using merchant data and other sources. I want the feed and website telling the same accurate story, not a retailer neglecting one to chase visibility in the other.

There is a sensible commercial reason to improve product information. A shopper looking for a particular material, size or delivery window needs a clear answer before buying. But filling every blank field is not the same as helping that shopper choose. Google Merchant Center distinguishes required, conditional and optional attributes. I'd prioritise the applicable requirements, then the details customers actually need, rather than asking the team to populate everything because an article suggests any gap sends the sale to a competitor.

The budget conversation needs more care too. If someone proposes a separate AI product feed optimisation budget, what exactly are we buying that the existing catalogue work does not cover? OpenAI supports applications for direct feeds, but that is not a promise of inclusion. Separately, Google says supporting links in AI Mode and AI Overviews need no special schema. Neither is a reason to ignore genuine data gaps, but an AI label on the proposal is not enough to justify extra spend.

Accurate product information is worth investing in even without a new traffic source. Where AI referrals can be identified, I'd follow them through analytics tools and order reporting to see whether they bring worthwhile sales. A stronger CVR could simply mean those shoppers arrive closer to purchase. It would not, on its own, prove that the feed changes generated extra business. Keep improving the catalogue, but do not build the sales forecast around recommendations nobody has guaranteed.