We’re writing this policy now. Until it’s published, here is how the product is built, which won’t change:
- SDKs evaluate flags in your process. The attributes of the contexts you evaluate are never sent to featflags.
- Evaluation counters carry the flag, variant, reason, rule and the version of your flags that served it, and no user data.
- Exposure events go to a callback in your code; we don’t receive or store them, except for flags in an experiment you run (below).
- Experiment data. When you run an experiment, your applications also send us, for that experiment only: a pseudonymous unit id, computed from the user’s key with a secret key that only you hold; which variant was shown; and the name and number of each event you chose to track (for example
checkout.completed, 49.90), timestamped to the minute. We can’t link a unit id to a person, and the same person has unrelated ids in different experiments, so we can’t link experiments together either. Nothing else about the person, device or request is stored with it. Per-unit rows are deleted 30 days after the experiment stops; the results we keep are totals per variant. - To sign you in, we keep your email address and the identity your Google or GitHub account returns.
- This site uses no cookies, trackers or third-party scripts.