The Programmatic Glossary

A practical glossary designed to help marketers, advertisers and ad-tech professionals understand the terminology that powers programmatic advertising

Back to the glossary
Fraud & Brand Safety
Traffic Scoring

An AI (Artificial Intelligence)-driven process of evaluating the quality and likely value of incoming ad requests in real time, assigning a score to each bid request based on how likely it is to be genuine, engaged human traffic versus fraudulent or low-value inventory.

HOW IT WORKS

Machine learning models analyse dozens of signals in each bid request: IP (Internet Protocol) reputation, user agent patterns, browsing velocity, device fingerprint consistency and geographic anomalies. Each request receives a quality score. High-scoring requests are routed to premium demand and priced accordingly; low-scoring requests may be filtered out or sent to lower-value demand.

EXAMPLE

An SSP's (Supply-Side Platform) traffic scoring system flags a spike of impressions from a cluster of IP (Internet Protocol) addresses in a data centre, all browsing at identical speeds and generating identical click patterns. These are assigned a near-zero quality score and excluded from auctions. Advertisers are protected and the publisher's inventory quality score improves.

RELATED TERMS

Invalid Traffic (IVT) | Bot Traffic | Ad Fraud | Brand Safety | Dynamic Floor Pricing
★ LIMELIGHT INSIGHT: Limelight uses traffic scoring as part of its supply quality controls, helping to protect advertiser budgets and maintain premium inventory standards across the platform.