Identity
Does every product have an id — and is every id unique? Duplicate ids make channels reject or silently overwrite items.
Free tool
Product Feed Inspector
Paste a feed URL and get a transparent health report — field coverage, defects, segment labels and a 0-100 feed score. Feeds up to 100 MB are streamed in a single pass.
What it checks
Six reads, one honest score
Each signal is a simple coverage and validity measure with a published weight — no black box.
Content
Title and description coverage. Empty or duplicated titles keep products out of search matching and lower ad relevance.
Media
Image coverage and https delivery. An item without an image cannot serve as a shopping or catalog ad at all.
Pricing
Price presence and validity, plus sale-price sanity: a sale_price at or above price breaks discount rendering everywhere.
Commerce data
Availability, landing links, brand and GTIN/MPN identifiers — the fields channels use to match, dedupe and enrich your products.
Segment labels
custom_label_0…4 presence and vocabulary — the columns campaigns segment on. The report shows which labels exist and what values they carry.
Why feed health decides ad performance
Every shopping and catalog ad you run — on Google, Meta, TikTok, Snapchat or Pinterest — is generated from your product feed. Before any bidding strategy or creative decision matters, the channel has to accept the item: a product with a missing image, an empty title, a duplicate id or an unparseable price is disapproved or dropped before it ever enters an auction.
The items that do get through still compete on their data. Titles feed search matching, images become the ad itself, availability and price drive freshness signals, and GTIN/brand identifiers decide whether the channel can match your product to its catalog. Weak fields mean fewer impressions on the same budget — which is why feed health is the first thing to fix, not the last.
This inspector reads your feed the way a channel does: it streams the XML or CSV record by record — feeds up to 100 MB in a single pass — measures per-field coverage, flags structural defects and rolls it into one transparent score. If a feed runs past the streaming window, the report says exactly how many items were analyzed instead of pretending it saw everything. The score is a technical signal of data health, not a prediction of ROAS — no tool can honestly promise that from a feed file alone.