The problem
With third-party sellers, the risk is not content volume but the specific content that should not be live: misleading listings, manipulated reviews, and questions exposing personal data. The internal team reviewed by sampling and reached the important cases late.
During campaigns, volume multiplies and the moderation queue becomes the bottleneck for the whole operation: a held listing is a sale that does not happen.
What we did
Corpshore took on moderation end to end, with written guidelines per content type and an immediate escalation protocol for the sensitive categories, which are what justify the whole operation.
Capacity was planned against the client's commercial calendar rather than the annual average: the team scales up six weeks before Black Friday and returns to base size in January.
The team also provides a monthly report of detected patterns, which the client uses to tune its automated rules and reduce the volume reaching human review.
Team shape
Nine moderators at base and up to twenty-four during campaigns, with task rotation and a wellbeing protocol for staff reviewing sensitive content.
Timeline
A 30-day pilot on one product category, extension to the full catalogue in month two, and the first full campaign covered in month five.
Data handling and compliance
Moderation covers published content and seller data, with standard contractual clauses for processing outside the European Economic Area. Buyer questions containing personal data are handled under the minimisation procedure agreed with the client.
Results
- -58 %
- Average time in the moderation queue
- 96 %
- Sensitive content escalated within 1 h
- -34 %
- Cost of the Black Friday campaign
- -22 %
- Volume reaching human review
Measured during campaigns, when delay costs money.
Against internal sampling that could take days.
Against hiring and training temporary reinforcement in Spain.
After feeding detected patterns into the automated rules.
Figures relate to the period stated in each case and depend on each client's starting point.
The monthly pattern report ended up being more valuable than the moderation itself. It let us fix the problem upstream.
