Fresh coffee, ground in seconds
Conical burrs, 40 grind settings and a quiet motor.
Feature flags and A/B testing
A guard gives each visitor a random id in a signed cookie. A hash of this id decides which variant of the experiment the visitor sees and which features are turned on. As everything happens on the server, the page doesn't flicker like with client-side testing tools, and visitors without JavaScript take part as well.
New: our grinders now come with a two-year warranty!
Your assignment
Visitor id 9RdXeeYxjYL-pzCmhTHwbg sees variant B. Start as new visitor to get a new id, and maybe another variant.
| Flag | Rollout | Your bucket | State | Preview |
|---|---|---|---|---|
bannerShows a banner about a new feature. | 50% | 0 | on | Turn on |
reviewsShows customer reviews below the product. | 20% | 31 | off | Turn on |
checkoutA new checkout which isn't finished yet. | 0% | 56 | off | Turn on |
A flag is on if your bucket (0 to 99) is below its rollout. Raising the rollout from 20% to 50% keeps it on for everyone who had it before. Preview links turn a flag on for one request, e.g. for testers.
Results
| Variant | Headline | Visitors | Bought | Rate |
|---|---|---|---|---|
| A | A coffee grinder for every kitchen | 0 | 0 | 0.0% |
| B | Fresh coffee, ground in seconds | 1 | 0 | 0.0% |
Each visitor counts once per column. The counts live in memory and start over when the server restarts.