AI
Multi-armed Bandit
A multi‑armed bandit is an adaptive way to split traffic among versions during a test. It gives more visitors to the options that are doing well, while still trying the others enough to keep learning. This can increase overall conversions while the test is running. It is useful when you want to make progress and learn at the same time. Example: Start with equal traffic to three page versions, then let the system shift toward the one with more add‑to‑carts.
Read the full guide
Go to our complete guide to learn everything you need about
Multi-armed Bandit
