810 Disapprovals During a Three-Week Promotion
An electronics merchant has 6,800 SKUs in Merchant Center. Two weeks into a three-week promotion, 810 items are disapproved: 520 for 'Price mismatch' and 290 for 'Availability mismatch'. The website shows a member price, 8% below list, to logged-in shoppers, while the feed sends list price. Site prices update in real time; the feed updates every 24 hours. Separately, 240 of the availability disapprovals are items that are out of stock for shipping but in stock for same-day pickup at 60 stores; the feed marks them 'out of stock'. Conversion value on the disapproved items fell 14% over the period, and campaign ROAS dropped from 480% to 410%. The feed team proposes switching to a 6-hour update cadence and pausing the promotion.
1. Order the investigation and fixes for the 520 price-mismatch items. State what the landing page, structured data, and feed must agree on.
Read the worked answer
Order
- Establish the price Google actually sees. Fetch a sample of affected product pages logged out and from a clean user agent. If the page shows list price to anonymous visitors and member price only after login, the mismatch is real for the crawler and comes from the site, not the feed timing. If the anonymous page shows member price, the feed is the culprit.
- Check what the promotion actually changed. A promo start that only edited the site would produce a step-change on day one; check the disapproval timestamps against the promo start. A 24-hour feed lag can explain a two-week rollout, but not a single cliff.
- Check structured data on the same sample pages: the schema.org Offer price and availability must equal the visible price for an anonymous shopper.
- Fix at the source. If member price is login-gated, do not send it as the feed price - send the price a new visitor pays. If the promotion applies to everyone, send it as
sale_pricewithsale_price_effective_datematching the promo window, via a feed rule rather than hand edits. - Stopgap: enable automatic item updates for price and availability while the fix ships, so mismatches self-correct instead of accumulating.
Agreement required: anonymous landing-page price = structured data price = feed price (or active sale_price), and currency and tax treatment consistent across all three. Tightening the update cadence helps only if the underlying value is right.
2. Decide how to represent the 240 shipping-unavailable but pickup-available SKUs, and quantify the risk of pausing instead.
Read the worked answer
Representation
These are two different promises and should be handled separately:
- For the standard Shopping feed,
availabilityshould describe whether the item can be bought online. If the item can be ordered online for pickup, it is purchasable - keep it in stock and make the pickup option visible on the landing page so the promise holds. If it cannot be transacted online at all, mark it out of stock in the main feed; that is accurate and avoids an availability mismatch. - Use a local product inventory feed (store code, price, availability per store) to surface the 60 stores where pickup is available via local inventory ads. That is the mechanism built for store-level stock, rather than stretching the online availability field.
- Do not leave the feed saying 'out of stock' while the landing page offers a purchase path; that is the exact mismatch generating the disapproval.
Risk of pausing
Pausing solves the symptom and removes the demand. The 810 disapproved items represent roughly 12% of the catalog; with conversion value already down 14% and ROAS down 70 points, pausing throws away the recovery and resets learning on those SKUs. The cheaper sequencing is: automatic item updates plus accurate values first, then re-review in 48–72 hours. Also model the loss: at the observed post-disapproval run rate, three more weeks of the promotion carries most of the remaining risk, so fixing within days recovers more than pausing ever will.