Using product tags for the way shoppers actually ask
Tags are organisation data AI platforms consider. The trap is that most catalogs use them for operations rather than for shoppers, filling them with codes like bfcm-eligible that describe a promotion rather than the product.
In short
- The product stays in Shopify Catalog, but assistants have less to match against when a shopper describes what they want.
- Found in 26% of the catalogs we scanned.
- This one needs a human decision, so Endcap finds and groups it but does not write it for you.
Why this weakens matching
Tags are organization data AI platforms consider when matching products to a request.
Tags are named by Shopify as product organization data that AI platforms consider.
How common it is
This appeared in 26% of the stores we scanned, affecting 0.5% of all products scanned. These are brands with staff and agencies, so it is not a small-merchant problem.
Most tag fields are operational junk drawers
Tags are the field merchants use for everything they could not put anywhere else. In the catalogs we scanned they held promotion codes, fulfilment rules, migration markers and internal states: bfcm-eligible, emp-disc-yes, GLS 2.0. None of that describes the product to a shopper, and all of it is technically a tag.
This is why we refuse to generate buying questions from tags, having tried it. A question set derived from operational tags reads as machine-produced nonsense and undermines the credibility of everything around it. If you want tags to help, treat them as a second descriptive surface rather than as storage: material, occasion, fit, use case, the words a shopper would actually say. And leave the operational ones alone, because your fulfilment process depends on them.
How to check it yourself
- Review your tag list and separate operational tags from descriptive ones.
- Add tags for attributes shoppers say out loud: material, occasion, fit, use case.
The fix
Add tags describing attributes shoppers actually ask for, such as material, occasion or fit.
How Endcap handles this
Endcap finds and groups every product with this problem and links each finding to the Shopify document behind it, but it does not write this one for you. Across the catalogs we scanned this affected 0.5% of products, which is the kind of number that is tedious by hand and quick in bulk.
The fix touches a field we deliberately do not automate. Endcap only ever writes vendor, product type and description, because those are the three where a wrong value is visible and reversible. Prices, identifiers, publication state and status are decisions with consequences a tool should not be making on your behalf.
Common questions
My tags are full of operational codes. Should I delete them?
No. Your fulfilment and promotion workflows probably depend on them. Add descriptive tags alongside rather than replacing what is there.
Do tags actually affect AI visibility?
Shopify names tags among the organisation fields AI platforms consider, so they carry some weight. They are not a substitute for a good description, and operational tags carry no shopper-facing signal at all.
Why does Endcap not generate buying questions from my tags?
We tried and removed it. Tag fields hold things like bfcm-eligible and GLS 2.0, and turning those into questions produced output that read as machine-generated nonsense and undermined everything around it.
Source: Shopify documentation
Check your own store
Enter your Shopify URL and see whether the public products we can inspect show these problems. No install required for the first scan.