AI describes a category one way. The OEM describes the same category another way. The distributor names it a third way. The aggregator parametric tree splits it across two. None of them are wrong. All of them are different.
That gap is the new sourcing landscape. It used to live underneath the catalog, invisible — a problem for the buyer who already knew what they were looking for, and a problem nobody else worried about. AI made it visible. Buyers now start their search by asking a model. The model uses its own vocabulary. The catalog uses different vocabulary. The brand uses a third. The buyer never sees the same shelf the seller operates on.
For non-inventory brokers, sourcing desks, and category buyers, this is an arbitrage. The categories where the funnel disagrees most are the categories where the brands AI recommends are not the brands the distributor stocks, and the brands the distributor stocks are not the brands the OEM markets. Whoever bridges the vocabulary, captures the margin.
For OEMs and product teams, this is a risk you can't see in your existing analytics. Your product page taxonomy can be perfect by your own internal standards and still drop your brand from the AI shelf buyers see first. Tracksuit and Latana measure brand awareness. Semrush measures search rank. Citation trackers measure mentions. None of them measure the gap between how AI categorizes you and how your channel does.
CategoryRank measures the gap. Category by category, week by week, with the math behind the disagreement made transparent. Not visibility scores. Not citation counts. Vocabulary fragmentation, the actual structural edge of the funnel itself.
Sophisticated sourcing teams already know the funnel is broken. We're building the surface that proves it — and surfaces the arbitrage where the brokerage margin still lives.
Select categories are unlocked publicly. The rest open as your wallet grows.