Canary Testing

Canary testing rolls a change out to a small group of users first, then expands it bit by bit. This limits risk because only a few people see the new code at the start. You can watch key measures like errors, speed, and conversions during each step and, if things look healthy, increase the share of traffic. If not, you roll back.

Example: You release a new search service to 2% of visitors for an hour and monitor usage. Then you raise to 20% if error rates stay low, steadily increasing toward 100%.

‍