ConfigCat

Does every flag check download a new setting?

The ConfigCat library downloads a file of settings and checks flags against the copy it has. Opening two screens can involve two flag checks and no additional downloads if the library already has the file.

Automatic polling fetches updates on a schedule. Lazy loading can fetch when a check needs fresh settings. Manual polling leaves refresh timing to the app. That choice affects how soon a changed flag reaches the person using it.

Which value appears when a flag rule does not match?

The dashboard’s final value applies when the library can evaluate the settings but no earlier rule returns a result. The app supplies a separate fallback for a check that cannot get a valid result.

How do you check whether a setting updates when expected?

Use a disposable copy and verify that its ConfigCat connection is for a separate test environment. Before opening the copy, confirm the rule targets only fictional testers and the app uses test-only records, notifications and event destinations. If you cannot verify the connections, inspect the refresh settings and trace the expected values on paper.

1. 1
   Choose two messages

   In the test configuration use “First practice message.” In the app set the fallback to “Practice fallback.” Check the first message with the test configuration available.
2. 2
   Change only the test setting

   Change the test value to “Second practice message.” Wait for automatic polling, request the flag after the configured cache interval in lazy mode, or use the copy’s manual refresh action. Then check what the app displays.
3. 3
   Check the missing-result path separately

   Use an empty-cache test instance configured to stay offline, with no network or shared cache for the flag. Confirm that it shows the built-in fallback. Keep the earlier refresh check separate from this one.

What does ConfigCat provide for an A/B test?

Percentage options can divide people into groups that receive different values. A variation ID identifies the value returned. To compare results, send the relevant version and outcome to an analytics setup that can match them.

An app might check a banner flag even if the visitor never sees the banner. Check that you record the version at the moment that matters and can connect it to an outcome such as completing the next step.

What counts toward the free plan and what can you keep?

ConfigCat lists five million configuration downloads and 20 GB of traffic per month on Free, with ten flags or settings per product. Reads from the saved copy are separate from the download limit.

ConfigCat documents a JSON export of the current product settings. Keep the rules, identifier choices and app fallback with it. Check that you can reproduce the expected decisions in a test copy before switching systems.