What an item-level issue means
Merchant Center sorts problems by how far they reach. An item-level issue belongs to specific products: one row, a handful of rows, or a whole slice of the catalogue that shares a fault. The rest of the feed keeps serving normally. That is the entire distinction — the same rule broken account-wide is reported differently and stops everything at once.
Each issue carries a severity. Some block the product outright, so it disappears from shopping surfaces altogether. Some limit it, meaning it can still serve but not everywhere or not with every feature. Some are only warnings: the product runs, but Google is telling you the data is weaker than it could be and matching will suffer for it. Reading severity before you read volume is what stops you spending a morning on warnings while blocked best-sellers sit idle.
Why item-level issues matter
Because they are quiet. Nothing in your campaign reporting announces that a group of products has been blocked. You watch impressions or revenue drift downwards and reach for bids or budget, when the real cause is that part of the catalogue simply stopped being eligible. On a large feed this can run unnoticed for weeks.
They also cluster. Item-level issues rarely scatter at random; they gather around a product type, a particular supplier’s data, or a template that produces the same malformed field every time. That is good news, because one fix upstream usually clears a long list in one go.
Where item-level issues go wrong
The most common handling error is triage by count. A list sorted by how many products are affected puts a harmless warning at the top and a block on your highest-margin line somewhere near the bottom. Sort by severity first, then by what the affected products are actually worth to the business.
The second error is fixing symptoms. Editing a value in a supplemental feed clears the flag but leaves the source producing the same fault, so it reappears with the next full refresh. If an issue keeps returning, the fix belongs in the shop, the export or the mapping — not in the patch laid over the top of it.
What to do about it
Work down the list in order of severity, then by revenue behind the affected rows. For each cluster, look for the common attribute rather than the common product, because that is where the cause sits. Confirm the fix by watching the issue count fall after the next refresh, not by assuming the edit took hold. A feed that generates the same faults month after month usually needs its structure rebuilt rather than patched, which is the substance of Merchant Center feed management. If a problem turns out to reach every product at once, you are looking at an account-level issue instead, and that is handled quite differently.