Skip to content

Create a Local Evaluation Flag

Guide to creating a local evaluation flag, or migrating an existing remote evaluation flag to local evaluation.

Create a local evaluation flag or experiment

  1. Create a flag or experiment as you normally would, but select the Local option for the Evaluation Mode setting (if you cant edit this setting contact us and we can turn on the feature).


  2. Within your new experiment, set your server-side deployment, configure the allocation (for example, target 100% of users), and activate your flag.



Your local evaluation experiment has been set up and is ready to using a local evaluation sdk.

Migrate an existing remote flag to local evaluation

You may retroactively transition a remote flag to local evaluation mode by changing the flag settings and updating the bucketing key from Amplitude ID to Device ID or User ID.


Migrating a flag from remote to local evaluation mode will change the bucketing behavior. This is not recommended for active experiments. You need to make sure that all targeting rules abide by the limitations of local evaluation.

  1. Change the flag's "Evaluation Mode" setting to "Local"



  2. Update the "bucketed by" setting from Amplitude ID to Device ID in each "Rule Based User Segment" and "All Non-Targeted Users" Section



    If you don't see the "bucketed by" setting please contact us to turn on the feature.

Last update: 2022-08-01