Scoring
- 0 — broken or missing. The line fails, or you cannot prove it passes.
- 1 — present but improvable. It exists and mostly works, with known gaps.
- 2 — strong. It passes on a sample of ten items, not just on one.
Loading…
Free resource
How to score
Score a sample, not the whole catalogue: ten items picked at random, plus the three best sellers. If a line fails on the sample, it fails on the catalogue.
Group 1
Titles and descriptions are your own writing. Supplier copy that names a different brand or manufacturer is a misrepresentation signal, not a shortcut.
Group 2
A reviewer compares the image to the title and the price. A mismatch is a data failure, even when the feed itself is valid.
Group 3
Identifiers are the line most merchants get wrong, and the one Google checks hardest. An invented GTIN is worse than a missing one.
Group 4
Price and availability are checked item by item, not on average. One mismatched item in ten is a 10% defect rate, not a rounding error.
Group 5
Structure is what keeps a clean feed clean. A feed that is fixed by hand once a week is a feed that breaks once a week.
Group 6 — folded in from the Product Data Quality Scorecard
Data quality is the folded-in half of this scorecard: the five dimensions that decide whether the feed can be trusted at all. A feed can pass every optimisation line and still fail here — and if it fails here, the optimisation does not matter.
Read the score
The score is comparable month to month, which is the point. A number that falls is a process breaking; a number that rises is a fix that held.
Related
Keep going
Send the details and we will tell you what to fix first, or quote the work if you would rather hand it over.
Prefer email? Write to md@fhavi.com