Definition
An optimization technique that allocates more traffic to better-performing variations while still testing others, rather than splitting traffic evenly. It's like A/B testing's smarter cousin who actually uses the results in real-time.
Example Usage
We ran a bandit test that shifted 80% of traffic to the winning variant within two weeks.
Origin
From the multi-armed bandit problem in probability and machine learning theory.
Fun Fact
Named after the thought experiment of a gambler choosing between multiple slot machines with unknown payouts.
Source: Data science and optimization marketing terminology
Related Terms
Translate This Term
See “Bandit Testing” in Corporate Speak, Gen-Z Slang, Pirate Speak, and more.
Try the Translator