Why WooCommerce AI Discovery Still Gets Price and Stock Wrong

Before choosing a plugin, draw the fact chain for one product

Teams often phrase the problem as “AI cannot find my WooCommerce product.” Separate at least six questions: does discovery resolve the right URL; are title, description and images readable; do price and currency agree; do stock and sellable status agree; can a variant be selected; and does the promise still hold at cart and checkout? Treating these as one “AI indexing” switch creates false diagnoses.

WooCommerce’s Store API is an unauthenticated JSON interface for customer-facing product, cart and checkout functionality, but it reflects the current customer session rather than every private backend record. The WordPress REST API is a general content interface. For AI, visible page content and machine-readable endpoints are two paths to maintain.

What WooCommerce provides—and what it does not guarantee

The official Store API docs list /wp-json/wc/store/v1/products together with cart and checkout endpoints. Products can be filtered by slug, SKU, price, stock status and catalog visibility; unpublished products are excluded. It is useful for a frontend, catalog reader or controlled retrieval layer, not a distribution channel that guarantees AI traffic.

Extra facts such as voltage, box contents, warranty region or compliance labels can be added through Store API extensions. In WordPress plus WooCommerce, keep the layers distinct: stories and FAQs can use the WP REST API, while product, cart and checkout facts belong to WooCommerce interfaces and pages.

Take a read-only snapshot: compare page, Store API and markup

Choose one sellable SKU and record its URL, product and variant IDs, capture time, visitor country and currency. Inspect visible price and stock, request /wp-json/wc/store/v1/products?slug=your-slug, then inspect Product/Offer JSON-LD in the initial HTML. If caching is used, record response headers and cache age.

Do not stop at “there is a price field.” Compare active price, currency, availability and variant. Google Merchant Center requires price and availability to match the landing page, checkout and structured data; Google recommends Product data in the initial HTML and testing with Rich Results Test or Search Console. These help systems understand facts but do not guarantee a rich result or AI recommendation.

Worked example: a US filter page is found, but the wrong variant is recommended

The following “Northline Flow” example is fictional. The Starter 2-pack is 39.00 USD and the Family 4-pack is 69.00 USD. The page defaults to Starter; the Store API returns both variants, but JSON-LD exposes only the parent’s 39.00 USD. A retrieval system sees the 4-pack title but receives the parent price, so the shopper finds the mismatch on the page.

Fix it in four steps: map variant names and SKUs across page, Store API and markup; use the current market’s currency and price for each Offer; expose stock at variant level; and state that price is confirmed for the selected variant at checkout. Run six queries—brand plus category, use case plus category, both variants, an out-of-stock variant and US delivery—and record URL, variant ID, price, availability and checkout result.

If page and API agree but an AI system still does not mention the product, the honest conclusion is “the fact chain is ready; discovery or recommendation has not been observed.” Do not respond by stuffing keywords or inventing reviews. Check Search Console, crawlability, catalog submission and channel eligibility next—not the product title again.

Turn updates into an ownership table, and stop in the right order

Assign an owner to each fact: product owns title, specs and use cases; inventory owns stock_status, backorders and locations; commerce owns price, promotions and currency; engineering owns JSON-LD, Store API extensions and cache invalidation; support owns delivery, returns and warranty boundaries. Keep at least captured_at, product_id, variant_id, market, page_price, api_price, markup_price, stock_status and checkout_result.

If the Store API returns 404, confirm that the product is published, the slug is correct and catalog visibility allows it to appear. For price mismatches, check market/currency, sale dates and cache. For stock mismatches, check parent versus variation stock management, backorders and locations. If only an AI summary is wrong, preserve page and API evidence and correct the external description manually. Without a test account, do not claim checkout verification; without permission, do not request or copy administrator tokens.

This method can show whether page and public interfaces contradict one another; it cannot prove that Google, ChatGPT or another model will index, recommend or convert the product. Separate crawlable, indexed, recommended and key-event states so the next action and owner are clear.