The missing product facts agents need: GTIN, shipping and returns

When people talk about "agent-ready" product pages, they often mean better storytelling. Blend Commerce's September audit of 200 UK and US retailers suggests the gap is more basic than that.

Blend could read product pages at 171 of the 200 stores. Most of those pages had the basics. The detail is where they fell down.

What Blend actually found on readable pages

Show the data

Field	Value
Product schema	82%
Price	76%
Availability	73%
Shipping details	22%
Returns policy	20%
GTIN barcode	18%

Source: Blend Commerce, AI Shopping Audit 2026. Counts in the report: schema 141, price 130, availability 124, shipping 38, returns 34, GTIN 31.

Field	Retailers with field	Share of 171
Product schema	141	82%
Price	130	76%
Availability	124	73%
Shipping details	38	22%
Returns policy	34	20%
GTIN barcode number	31	18%

Source: Blend Commerce, AI Shopping Audit 2026.

Put the other way: 41 of the 171 were missing price from product data, and 47 were missing availability. Even the "basics" are not universal.

Why GTIN, shipping and returns matter to agents

A GTIN is the barcode number that identifies a product across retailers. Blend's explanation is the right one for CPG: without it, an agent comparing prices has to guess whether two listings are the same thing. The audit also notes that none of the beauty, fashion, luxury, pet or department store retailers whose pages it could read had a GTIN in that data.

Shipping and returns are part of the buying decision. If they are missing from product data, the agent has to hunt a policy page or leave them out. Shoppers do the same thing, only slower.

Discovery was healthier than detail: 107 of the 200 retailers published a product sitemap Blend could find, leaving 93 without one.

Shopify is ahead on checkout, not on barcodes

Shopify stores in the audit averaged 91 out of 100. Everyone else averaged 59. Thirty-seven of the 38 live agent checkout endpoints sat on Shopify (the other was Albee Baby). That endpoint gap is real. It is also not the whole story.

Field	Shopify stores with field	Share of 37
Product schema	36	97%
Availability	34	92%
Price	33	89%
GTIN	12	32%
Shipping details	11	30%
Returns policy	10	27%

Source: Blend Commerce.

Thirty-three of the 37 Shopify stores were missing at least one of GTIN, shipping or returns. Nineteen were missing all three. Blend traces this to platform defaults: Shopify's built-in structured data covers name, brand, price and availability, but Shopify's own documented example does not include GTIN, shipping or returns, so those fields appear only when a theme or app adds them. The four stores that scored 100 (Tanners Wines, Natural Baby Shower, Babythingz and Tiger Fitness) had the full set.

If you are on Shopify and feel "agent-ready" because Agentic sales channels are on, check the barcode field before you celebrate.

A CPG schema checklist

Work this list in order. Do not start with poetry.

1. Confirm the page is reachable. If bot protection blanks the PDP for automated visitors, schema never gets read. See our Blend think piece and tip 47.
2. Name, brand, price, availability. These should already be in on-page product structured data. If Blend's wider sample is any guide, do not assume they are.
3. GTIN on every variant. UPC, EAN or ISBN in the admin is not enough. The value must appear in the schema a crawler fetches. For multipacks, use the correct code for the sellable unit.
4. Shipping details in product data. Carrier expectations, regions, thresholds or "ships from" facts that affect the buy decision. Link out only as a backup, not as the only copy.
5. Returns policy in product data. Window, condition and who pays return shipping, in the structured data and in human-readable text.
6. Product sitemap. Reference it from robots.txt. Blend could not find one for 93 of 200 retailers.
7. Validate. Run a live PDP through Google's product structured data documentation guidance and Rich Results Test. Fix errors before the next catalog push.
8. Match the feed. Merchant Center or marketplace attributes should not contradict on-page schema. Agents and shopping modules both notice drift.

All readable (171)Shopify (37)

Show the data

Field	All readable (171)	Shopify (37)
GTIN	18%	32%
Shipping	22%	30%
Returns	20%	27%

Source: Blend Commerce, AI Shopping Audit 2026. Shopify shares are higher but still leave most stores missing at least one field.

How this differs from "agents fail on product facts" stories

We have covered agent failures that come from vague claims and missing facts on journeys (CartTrace-style work). Blend's contribution is a shelf-level frequency count: how often the structured fields are simply absent across 171 readable stores. You do not need a failed journey log to justify adding a GTIN. The base rate is enough.

PPC Land's November 5 write-up is also clear that Blend sells conversion audits and discloses its Shopify interest. Read the shares as a snapshot with a commercial author, not as a census of all retail. Ten retailers per sector is a small group. The direction of the gap still matches what we see on CPG catalogs: barcodes and policy fields lag hero copy.

A one-sprint plan for a brand team

Days 1-2: Export your top 50 SKUs. Note which lack GTIN, shipping schema and returns schema on the live PDP (fetch HTML, do not trust the admin alone).

Days 3-4: Fill barcodes. Confirm theme or app output. Fix the ten worst PDPs.

Days 5-6: Add shipping and returns properties to product structured data templates, not as a one-off paste on a single page.

Day 7: Retest with Rich Results Test and one agent prompt per SKU ("compare this with a peer on price and returns"). Log what the agent still cannot answer.

Multipacks, cases and the wrong barcode

CPG catalogs break GTIN work in predictable ways. The sellable unit on Amazon may be a 12-pack while the brand site sells singles. The case code in the warehouse system is not the unit code on the carton a shopper receives. Agents that match "same product" across retailers need the code for the offer they are looking at, not the code for a different pack hierarchy.

Before you bulk-upload barcodes:

• Confirm whether each PDP sells an each, inner pack or case.
• Reject placeholder GTINs and reused codes across unrelated flavors.
• Retest after a theme update; schema apps sometimes drop the gtin property when a developer restyles the buy box.

Shipping and returns fields have a similar trap. A sitewide footer link is not the same as product-level structured data. Blend scored what it found in product data on the page, not what a determined human could eventually click.

XENA Intelligence looks at how assistants describe products when the facts are available to read. SearchPanel scores that visibility at SKU level. Neither tool can invent a barcode your page never publishes.

Where this sits in a Black Friday week

Peak season is when missing fields hurt most. An agent comparing two sunscreens, two snack packs or two detergent pods will favour the listing it can identify with a barcode and the one that states return windows without a side quest. Blend's snapshot is from mid September, not from Cyber Week, but catalog hygiene does not get easier once traffic spikes.

If you only have capacity for one field this week, ship GTINs on the variants that already sell. Shipping and returns templates can follow in the next sprint. If you only have capacity for a conversation, ask whoever owns the theme or feed whether barcode values ever leave the admin. That single question explains most of the Shopify gap Blend measured.

Keep the checklist next to the bot-protection allow-list question. A perfect schema on a page the agent cannot load is still a silent miss.

The short version

Agents compare. Comparison needs stable identifiers. Agents buy. Buying needs shipping and returns facts beside price and stock. Blend's 18%, 22% and 20% figures say most readable stores still skip those fields. Fill them. Then worry about the clever copy.