First, read the feed's last successful fetch and processing history
Open the data source for the affected products and read its processing history from the most recent run backwards. Note the time of the last run that succeeded, the time of the last attempt, and the item count each processed. A run of failed attempts after a certain date means the fetch broke. Successful runs that are simply too far apart for how often the store changes mean the schedule is the problem. Successful runs with the expected item count, while items still show old values, mean the data is arriving and something later is changing it, or the source itself is stale.
Then pick a handful of stale items and compare them with the store and the live product pages: the price and stock in the store's admin, on the page, in the most recent feed file, and in the item's details in Merchant Center, which show the value each layer contributed. Where those diverge is where freshness breaks. Keep this separate from items that do not appear in ads at all, which the shopping products not showing diagnosis covers, including disapprovals for price mismatch; this page is about data that is present but out of date.
Diagnosis table
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| Processing history shows failed fetches since a given date | The fetch error message; whether the file location, address or login changed | Scheduled fetch failing after a moved file, changed address or credentials | Point the fetch at the current location with valid credentials; fetch now |
| Fetches succeed, yet prices lag changes made during the day | Fetch frequency against how often prices and stock change | Fetch schedule slower than price and stock changes | Fetch more often, or move fast-changing fields to an API or integration |
| Items vanish or show as expired though still on sale | Item last-update dates; whether some items were dropped from recent files | Items expiring because they were not refreshed in time | Include every live item in every full feed; refresh items before they expire |
| No updates at all since a platform change, with no fetch errors shown | Integration connection status; API error logs; item update times | Store platform integration or API sync broken after a change | Reconnect or reauthorise the integration; repair the API job |
| Page shows the new price; Merchant Center keeps the old one between fetches | Automatic item updates setting; structured data on the product template | Automatic item updates off, so page changes are not picked up | Enable automatic item updates and fix the product markup |
| Primary feed correct, item still shows an old value | The item's attribute details by layer; supplemental feeds and rules | Supplemental feed or feed rules overriding fresh values | Remove or correct the override; document which layer owns the attribute |
Fix, verify and prevent each cause
A fetch that started failing
Scheduled fetches break for mundane reasons: the feed file moved when the site was rebuilt, the export address changed with a new feed app, the password on a protected location was rotated, or the file now exceeds what the fetch can read. Merchant Center keeps the last good data and records the failure, so the store looks fine until prices change.
Fix: read the error on the failed runs, update the fetch location and credentials, and trigger a manual fetch, following the feed-level diagnostics described in the product feeds guide; the Merchant Center help centre documents the fetch error messages.
Verify: the manual fetch succeeds with the expected item count, and the next scheduled run does the same.
Prevent: whenever the site, the feed app or the hosting changes, add "fetch location and login" to the change checklist.
A schedule slower than the store
A feed that is fetched once a day cannot keep up with a store that changes prices in the morning and sells out in the afternoon. The data is correct as of the last fetch, which on a busy day is wrong for most of it. Sale launches expose this most painfully, because prices change for many items at once.
Fix: fetch more often where the source allows, and for stores with changes during the day, move price and availability to the store platform's integration or the API, which the product feeds guide recommends for intraday changes and flash sales.
Verify: the gap between a price change in the store and the same price in Merchant Center shrinks to fit the business's rhythm, tested by timing a deliberate change.
Prevent: set the update frequency from how often the fastest-changing field changes, not from the platform's default.
Items that expired quietly
Merchant Center treats items as current only while they keep being refreshed; an item not updated within the window the help centre documents expires and stops serving. Items expire when a feed is sent as a partial file that omits unchanged products, when a filtered export leaves out slow sellers, or when an API job only sends items whose data changed.
Fix: make sure every live product is included in each full feed, or that the API job refreshes every item before it would expire, as the product feeds guide implies when it lists expired feeds among the feed-level issues that stop updates.
Verify: the expiring-items count in diagnostics falls to nothing over the following days, and item last-update dates are recent across the catalogue.
Prevent: monitor the item count in each processing run against the number of live products, and alert when it drops. Products that are genuinely discontinued are better removed from the feed than left to expire, as the guide to new, seasonal and clearance products explains alongside out-of-stock handling.
An integration that went quiet
Stores connected through a platform app or a custom API job do not have a fetch schedule to inspect, so a broken connection can look like a calm week. Reauthorising an app, changing the store's plan or theme, renaming a product field or a failed deployment of the job can each stop updates without any error appearing in the feed view.
Fix: check the integration's connection status in both the store platform and Merchant Center, reauthorise it if needed, and read the API job's logs for rejected requests; the product feeds guide names the platform integration and the API as the routes for stores with fast-changing data.
Verify: a test price change in the store reaches the item in Merchant Center through the integration, and item update times start moving again.
Prevent: alert on the age of the most recently updated item, not only on errors, so silence is noticed.
Automatic item updates switched off
With automatic item updates on, Merchant Center can read Product and Offer structured data on the landing page and correct the item's price and availability to match, which covers the hours between feed updates. If the setting is off, or the markup is missing or wrong on the product template, the item stays at the feed's value until the next update.
Fix: enable automatic item updates for price and availability, and make sure the product pages carry accurate structured data on the final URL that matches the visible price, as the product feeds guide explains in its section on the structured-data sync.
Verify: after a deliberate price change on a test item, Merchant Center reflects it before the next feed update.
Prevent: include structured data in the checks for any product template change, and remember the setting is a net beneath the feed, not a replacement for it.
A later layer overwriting the fresh value
Merchant Center reads the primary feed first, then applies supplemental feeds by item ID, then feed rules. A supplemental spreadsheet built months ago to fix titles that also carries a price column, or a rule that builds availability from a field the store no longer updates, will overwrite correct primary data on every run. The processing history looks healthy and the primary file is right, yet the item is wrong.
Fix: open the item's attribute details to see which layer set the stale value, then remove the column from the supplemental feed or correct the rule, as the product feeds guide describes in its section on supplemental feeds and rules.
Verify: after the next processing run, the item shows the primary feed's value.
Prevent: keep a short register of which layer owns each attribute, and allow price and availability to come only from the primary source.
Prevention: fetch monitoring and a sale-day schedule
Feed freshness fails silently, so prevention is mostly about making silence visible. Three alerts cover most of it: a failed fetch or upload, a processing run whose item count falls well below the number of live products, and an item last-update time older than the business would tolerate. Read the processing history in the weekly Merchant Center routine the product feeds guide sets out, and add the fetch location, login and integration status to the checklist for any site, theme, app or hosting change.
Sale periods need their own plan. Before a promotion, confirm the route that will carry the new prices, schedule additional fetches or confirm the integration pushes changes immediately, check that sale prices use the sale price attribute with effective dates rather than overwriting the regular price, and run a test change a day ahead. Stale prices on a sale day cost twice: shoppers click on an old price, paying a cost per click (CPCAverage cost of each click.) for a visit that disappoints, and items are disapproved during the hours that matter most. Because the cost falls hardest on high-value products, watch freshness first on the items with the highest average order value (AOVAverage revenue per order.), and report the return on ad spend (ROASRevenue generated per unit of ad spend.) of the sale period with any freshness incidents noted beside it.
A homeware store's prices stop moving
A homeware store with 2,600 products runs a summer sale starting on a Monday. By Tuesday, customers complain that Shopping ads show full prices. The primary feed's processing history shows successful daily fetches, each with 2,600 items, but the item details show the sale price arriving from the primary feed and then being replaced by the regular price from a supplemental spreadsheet built eight months earlier to fix titles, which also carried a price column. In addition, the last successful fetch was at 03:00 on Monday, before prices changed at 09:00. The team deletes the price column from the supplemental file, adds fetches at 09:30 and 15:00 for the sale fortnight, and turns on automatic item updates after fixing missing Offer markup on the product template. By Wednesday every sampled item shows the sale price, and click-through on Shopping recovers from 0.7% to 1.3%.
Common mistakes
Reading the last fetch attempt instead of the last success. Sending partial feeds and letting unchanged items expire. Relying on automatic item updates instead of a fresh feed. Leaving price columns in supplemental files built for something else. Changing a site or feed app without checking the fetch. Planning a sale's prices without planning how they reach the feed.
