Moderation for listings, reviews and seller content
In a marketplace, spam isn't the only problem. Catalog quality and trust in reviews matter just as much. We fit the rules to your product categories, languages and the cases that are critical for your business.
- Listings, reviews and seller content checked through one API.
- Suspicious cases go to a REVIEW queue, most urgent first.
- The same rules across countries, languages and sales channels.
What we check in a marketplace
Listing titles and descriptions, buyer reviews and messages from sellers. Problems look different in each: in listings it is spam, contact details that lead off the platform and claims that must not be made; in reviews, fake or paid reviews; in messages, attempts to move the deal off the site.
Why one list of banned words isn't enough
Categories carry different risk. In supplements you watch for claims of curing diseases, in electronics for descriptions that hint at a counterfeit, in classifieds for contact details hidden in the text. Add spelling tricks such as "v1agra" for "viagra" or digits in place of letters. A word list will not catch that, and a strict filter starts blocking honest sellers.
So we fit the rules to your categories, and tune the model to your content. Whether it works for your catalog, we check on a sample in a pilot, not on slides.
What a decision looks like
Every listing gets a decision: ALLOW (publish), REVIEW (check by hand) or BLOCK (stop it). The decision comes with a confidence score and a reason, so the moderator sees why the system reacted and does not read everything from scratch. Cases with low confidence go to REVIEW instead of an automatic block.
Where to start
The simplest start is a pilot on a sample of your listings. We describe how it works on a separate page.