The FeedScore has a new calculation method

If your FeedScore has moved over the past few weeks, your catalog isn't what changed — the way we score it is. Here's why, and what the new score lets you do.

What the old score didn't tell you

The old FeedScore aggregated fill rates: how many enriched listings, how many errors, how many mapped categories. It was readable, but it measured what was filled in, not what was good — let alone what was making you money.

Three blind spots, concretely. A title that existed but was mediocre counted as a perfect title. An error blocking distribution weighed as much as a cosmetic suggestion. And above all, a listing nobody sees weighed as much as one of your best-sellers. You got a number, but no starting point.

The new method

We now score every single product listing, one by one, on technical criteria: title quality, description quality, required and recommended attributes for the product's category, category mapping accuracy. Those product-level scores are then aggregated into a catalog score out of 100.

The most important change is right there: the aggregation is weighted by impressions. Products generating the most views count for more in your score. So your score describes the state of what you actually serve, not the average of your entire stock — dormant items included.

A useful side effect: the gap between the simple average and the weighted average is information in itself. If it's positive, your well-optimized products are the ones getting the most impressions. If it's negative, it's the opposite — and you have volume to recover on your most visible listings.

The score isn't the useful part

A score that doesn't tell you what to fix is just a thermometer. On the Optimization home screen you'll now find a list of priority actions, ranked by severity and then by the number of products affected:

• invalid URL, missing or invalid GTIN or MPN — anything blocking distribution always comes first;
• description too short — the item that hurts product understanding the most;
• titles too short or in all caps — quick fixes, immediate gain;
• missing attributes and overly generic categories — the groundwork, to handle in volume.

Each action tells you how many products are affected and links straight to the screen where you fix it: the Texts, Categories or Attributes tab, or the product list pre-filtered on critical cases.

And if your score drops below 30, the gauge turns red. Not to worry you, but to flag that a significant share of your catalog isn't working for you.

One clarification about AI enrichment

AI coverage — the share of your listings enriched automatically — is still tracked separately, on the Optimization page. It does not feed into the FeedScore calculation.

That sometimes surprises people: you enrich at scale, and the score barely moves. It's logical. Enrichment improves content; the FeedScore measures completeness and technical compliance. If the score doesn't climb after an enrichment run, it usually means required attributes are still empty, identifiers are missing, or categories are badly mapped — exactly what the priority action list now spells out.

The agent reads the same score you do

The Feedcast agent now relies on the same source as your own screens: the backend score out of 100, the impression-weighted version first, and the list of products affected by each improvement area. No more gap between what it says and what you see.

So you can ask it, in the chat: "Explain my FeedScore and the 3 most urgent actions." It reads your live data, prioritizes, and tells you what can be fixed inside Feedcast — as well as what has to be fixed on your store, because not everything is on us.

Why we keep pushing on this

Because listing quality is the most underestimated variable in ad performance. Google, Meta and the others read your product data before deciding what to show, to whom, and at what cost. A precise title, the right attributes, the right category: your product is better understood, so better served, so better clicked.

There's a broader ambition too. With more than 6,300 connected stores, we have what it takes to turn the FeedScore into a reference index for product listing quality in e-commerce: a score that means something, comparable from one catalog to the next, the way people talk about ROAS. We're not there yet. That's the direction.

In the meantime, your score is in the app, and the list of fixes that pay off most sits right below it. Learn more about the FeedScore.