Structuring Online and In-Store Inventory for Search and AI

Products that are out of stock online but in stock at a store are not unusual. Shops ask how to structure inventory information so that search and AI can tell people they can buy nearby.

The answer has two stages. First, represent the stock state separately for each channel. Don't mix online stock with store stock. Second, keep the stock wording visible on screen and the stock value in the structured data always the same. Structured data is a description of a page's content written in a form machines can read. When either of these breaks, search engines stop trusting the inventory information. AI drops the product from its recommendation candidates too.

This article covers how inventory information affects AI recommendations, a real case of hidden out-of-stock text, per-channel representation, the connection to store information, and update operations.

Conclusion: Separate by channel, and match the screen and the structured data values

For the structured data representation, use the Schema.org Offer and ItemAvailability vocabulary. What matters more than the choice of vocabulary is the match with the on-screen wording.

Key idea: AI treats a page whose on-screen wording and structured data values differ as untrustworthy information. In structuring inventory, consistency comes before precision. Compare the "On-screen display" and "Structured data representation" columns above against your own product pages.

How inventory information affects AI recommendations

When AI search recommends a product, whether it can be bought now is a main criterion. A product judged out of stock drops out of the candidates even with a good price and reviews.

The problem is that AI doesn't judge out-of-stock status from what's visible on screen alone. If template-derived "out of stock" text remains in the HTML, the AI reads that text even when an "Add to cart" button is showing on screen. Then it excludes the product. Check by searching the source of a product page that's on sale for the words "out of stock".

Case: Hidden out-of-stock text

This happened at a Korean ecommerce brand with annual revenue of about ₩40 billion (about ¥4 billion). Product pages for items on sale still carried "sold out" text that was only hidden, not deleted. To the human eye, they were normal product pages. But the AI may have read that text and dropped products that were on sale from its recommendation candidates.

The improvement began by removing this hidden text. The on-screen stock display and the structured data values were also brought into line. After that, product descriptions that existed only in images were converted to text. Watching fetch status in bot logs, product and category pages were improved first. As a result, organic purchase revenue grew about 80% in three months. The starting point of this improvement was consistency in how inventory was represented.

Per-channel representation: separate online from store

The basic structure is a separate Offer for online and for store on the same product. Even for the same sneaker, warehouse stock and the Shibuya store's stock are different shelves.

1. Online Offer: holds the price, availability, and shipping terms. availability refers to the online stock state.

2. Store Offer: for each store, holds availability and a reference to the store (availableAtOrFrom). With many stores, narrow it to main stores or a representative expression such as "available at stores".

3. On-screen display: shows the same content as the two above in wording people can read. Something like "Out of stock online. Available at the XX and YY stores".

Whether per-store stock can be shown on the product page depends on the platform's features. If it can't be shown, don't write per-store stock in the structured data either. Stay at the granularity of "carried in stores". Check that you aren't writing information in the structured data alone that isn't on screen.

Connecting to store information

For store stock to mean anything, search engines and AI first have to know about the store itself.

When product pages and store pages reference each other, AI can understand that this product can be bought at this store. Start by checking that store pages carry name, address, and opening hours as visible text.

Update operations: inventory moves

Inventory changes daily. If structured data updates lag, consistency breaks.

If lastmod updates on stock changes, crawlers stop trusting the update signal. Represent stock in structured data, and limit lastmod to content changes. Check that lastmod doesn't change when stock moves.

Operations for keeping consistency

Inventory structuring is assessed on today's state, not on the day it was designed. The more products, the more stores, and the more often stock moves, the heavier the load. Having people keep checking that screen and structured data match stops being realistic.

Search OS treats maintaining that consistency as an operation. It doesn't change the existing platform. It continuously verifies the match between visible text and structured data on product pages, and detects hidden stock text. It also watches the references to store pages and the operation of the sitemap and lastmod. Bot logs confirm whether search engines and AI crawlers are actually fetching product and store pages. On that basis, the places where discrepancies arise are compiled as fix candidates. You can start by picking one in-stock and one out-of-stock product and comparing the screen with the structured data.

Frequently asked questions

Should I write per-store stock counts in the structured data?

Stock counts (inventoryLevel) can be represented. But it's safer not to write information that isn't shown on screen. The more often a value changes, the higher the risk of a mismatch.

When online stock runs out, should I put the page on noindex?

It's safer not to. If there's store stock, the page has value as information that the product can be bought in store. Represent OutOfStock correctly, and place the store guidance as visible text.

Does inventory structured data appear in rich results?

Google's product rich results can show availability. But the focus of this article isn't rich results; it's consistency as a criterion when AI recommends products.

Related reading

• SEO/GEO Case Study: How an Ecommerce Brand with ₩40 Billion in Annual Revenue Grew Organic Purchase Revenue 80% in Three Months

• What Structured Data futureshop Outputs Automatically, and What to Add for AI Search

• 2026 Ecommerce Strategy: How to Plan for AI Shopping Search

References

• Google Search Central: Product structured data

• Google Search Central: Local business structured data

• Schema.org: Offer, ItemAvailability

• Google Merchant Center Help: Local inventory ads