Shopify Sends Compare-at Prices to AI Agents: Check the Price Pair

A product can have a sale price on its storefront while an AI shopping service sees only the current price. Shopify’s October 9 change addresses that missing context: Shopify Catalog now shares compare-at prices with agentic channels. Merchants can control that sharing through the Compare-at setting in admin.

The announcement does not name every receiving agent, specify country or plan limits, or guarantee that an agent will display a discount. The merchant can control the supplied data; the receiving experience remains a separate check.

A markdown is a relationship between two prices

The useful object to audit is a price pair for a given variant, currency, and market. Looking at the two numbers independently misses the relationship a shopper is meant to understand. A lower current price may represent a discount only if the comparison is meaningful for that same product and purchasing context.

In the first example, (100 - 80) / 100 = 20%. These are invented numbers to explain the calculation, not a store test or an AI-agent result. If one number is in euros and the other in dollars, or one belongs to a different variant, that calculation no longer describes the buyer’s offer.

Follow one product through three surfaces

Shopify’s help page gives the configuration path: Sales channels > Agentic > Sources > Shopify Catalog > Review product data mapping > Compare-at. Save the switch selection. The documentation says eligible comparison prices are sent only when higher than the current price, through the UCP Variant list_price field. The store begins with a default selection; later changes to comparison-price usage do not automatically change that selection.

The Catalog preview is useful for checking the configured pair, but Shopify explicitly says it is not an exact indicator of an AI channel’s display. Mapping changes can also take time to process. Do not convert a preview screenshot into a claim about a live agent answer.

Start in the merchant record. Select a product with variants and identify the exact option being reviewed. Note its market, currency, and current price. Then check the comparison value and the sharing setting. Shopify’s Catalog mapping documentation is the reference for how catalog data is configured; do not assume a theme label controls every downstream field.

Open the buyer-facing product page. Choose the same option and market. Read the price the customer can actually purchase at. If the page advertises one pair while the catalog record supplies another, repair the mismatch before measuring AI visibility. Screenshots from a different country or variant cannot settle it.

Inspect the receiving channel when accessible. Save the result, its destination URL, and capture time. A correct merchant record with a missing markdown in one answer may mean the channel did not use that field, the offer was not refreshed, or the product was not selected. It is not enough evidence to declare the sharing setting broken.

The failure cases worth catching

• A default variant’s comparison price is attached to an answer about a different size or bundle.
• An old promotion ends, but a reference value remains in one supplied record.
• A market-specific offer is checked against the wrong currency or storefront.
• A successful data handoff is reported as proof of agent display or increased sales.

These are review scenarios, not defects we observed in Shopify. Their value is that each has an actionable next check. For the variant case, compare product identity. For the expired promotion, compare timestamps and current records. For market mismatch, hold the locale constant. For the display claim, capture the actual receiving experience.

Treat better data as the first outcome

The immediate opportunity is clearer product information at a handoff, not a new documented ranking factor. Keep configuration success, transmitted data, visible offer, referral, and purchase in separate records. A price field can be correctly supplied without the product being recommended at all.

Our Shopify AI-checkout analysis explains why a downstream transaction is another boundary. If product text also changes across markets, the translation auto-sync report covers the freshness gap between a source edit and the translated copy. Review both paths before declaring an offer consistent worldwide.