Shopify Now Shows AI Agents Your Sale Prices. Four Settings to Check First

What changed

Shopify's changelog entry of October 9 is short: compare-at prices are now shared with AI shopping agents through Shopify Catalog. Until now an agent reading Catalog saw only the current price. The strikethrough price your storefront shows was invisible to it, so a product on sale looked like a product at full price. There is a toggle in the admin, under the Compare-at section of the Catalog settings, that controls whether agents see it.

Small entry, real consequence. Shopify Catalog is the feed that puts your products into ChatGPT, Microsoft Copilot and, through the respective sales channels, Google AI Mode, Gemini and Meta's surfaces. Shopify's help page on product discovery for agentic storefronts calls it the primary method for agentic storefronts to receive your product data, and says it continuously updates price and inventory across every AI channel. What Catalog sends is what the assistant knows about your product. This week it learned about your markdowns.

Why price signals matter more than they used to

Assistants do not just list products, they filter them. The clearest public example came from Meta's Muse shopping agent. Profound extracted the assistant's instruction set in late September, and PPC Land reported it on October 12: the agent is told to search at most three stores on the open web, return no more than ten products, rank well-known merchants first in its top five, favour direct sellers over resellers and discard listings priced far outside a category's normal range.

Meta has not commented, and the file may have changed since. Treat it as a glimpse, not a specification.

Even as a glimpse it says something merchants should hear. A price that reads as an outlier gets dropped before any shopper sees it. A sale price that the agent cannot see is a product that looks more expensive than it is. With the holiday window opening, that is the difference between being the deal in the answer and not being in the answer.

Four settings to review this week

1. Decide whether agents should see your compare-at prices

For most DTC brands the answer is yes. If you run real markdowns, you want the agent to describe the product as on sale, and Shopify's help page says agents may use the compare-at price to show strikethrough pricing. Say no if your compare-at values are not real former prices, if brand or MAP agreements restrict how discounts are displayed, or if you use compare-at as a placeholder on products that never sold at that number.

An agent quoting a fake "was" price is a trust problem, and in some markets a legal one. Clean the field or turn the toggle off, do not leave inaccurate values flowing to assistants.

2. Check what Catalog is actually sending

By default Catalog lists each product with title, description, options, images, price, availability and other key attributes. If your real titles, descriptions or categories live in metafields, metaobjects, tag prefixes or delimiters inside product titles, the default export will send the wrong field. Shopify Catalog Mapping exists for that case. Open it, pick ten priority products and compare what the mapping sends with what the product page shows.

The attributes shoppers ask assistants about, such as material, fit and dimensions, are the ones most often stuck in a custom field.

3. Understand that robots.txt does not stop Catalog

Shopify is explicit: blocking AI crawlers at the robots.txt or network layer affects only open-web discoverability and does not stop Catalog from sending product data to the agentic storefronts you have activated. Teams that blocked GPTBot in the summer and assumed their products were out of ChatGPT should check which channels are active. The reverse holds too. If you want products out of AI channels, Shopify has a separate procedure for hiding products, and the crawler block alone will not do it.

4. Read your own agents.md

Every Shopify store now serves /agents.md, plus /llms.txt and /llms-full.txt for older crawlers, with the store name, sitemap, policies and discovery endpoints. The content is generated by default, and you can override it with a Liquid template. Open yours and read it the way an agent would. If your return policy changed for the holidays and the file still describes the old one, that is what an assistant will tell a shopper.

Shopify also notes the discovery files are separate from Catalog and do not replace it, so this is about store context, not the product feed.

What to watch after you change the settings

Keep a fixed list of shopping questions and check how your sale products are described in ChatGPT and Google AI Mode before and after the compare-at change. Google's AI performance insights report in Merchant Center covers Google's surfaces only, so for ChatGPT the check is manual or through monitoring. Do not expect an overnight shift: Catalog updates continuously, but how each assistant uses the new field is its own decision.

This is the kind of change a Paz expert handles for lean ecommerce teams. We read what your systems actually send to each channel, including Catalog mappings and feed fields, find the products where the data disagrees with the page, fix them within guidelines you approve, and track how your products appear in AI answers over time. If you want to know what assistants see from your store before November, talk to us.