Catalog health and Quality Score
How Noobit measures whether a product is ready to be published.
The Quality Score is a 0-100 score Noobit calculates for each product. It reflects how complete and ready to publish the product is: data, content, images, metadata. Catalog health is the aggregate of all the individual scores in your account.
How it's calculated
The score starts at 100 and subtracts points for each problem it finds. There are more than 20 checks, grouped into six blocks:
Specific issues, not just a number
Each product doesn't just have a score; it has a list of active issues, each with a severity level:
Catalog health dashboard
In «Dashboard», «Salud del catálogo» (Catalog health) shows:
- The catalog's average score
- How many products are fine, have warnings or are critical
- Coverage: AI content, description, images, SEO, brand, EAN and published
- The products that need attention, with their main problem
Auto-approval threshold
Set per account in «Preferencias» → «Pipeline» (Preferences → Pipeline): the minimum score a product needs to be published automatically. It's 85 by default, but you adjust it to your level of confidence.
- Product with a score ≥ threshold: if the publisher is on Automatic, it publishes directly
- Product with a score < threshold: if the publisher is on Automatic, it needs manual approval even though it's set to Automatic
When it's calculated
Health is calculated on the spot every time you open the product. The quality score is saved when the product goes through quality control: in the pipeline or with the «Control de calidad» (Quality control) button in the product record (it reads «Re-evaluar QA», Re-run QA, once the product has a score).
What the score does NOT measure
The score deliberately doesn't try to measure the subjective quality of the content (whether the description “sounds good”) or the aesthetic quality of the images. It measures completeness and structural consistency. You judge the editorial quality by reading the product page; the look of the photos is checked by image processing, which sends the ones it doesn't accept to «Revisión imágenes» (Image review).

