Hi Team We have one experiment running for our users I was g GrowthBook Users #experimentation

Hi Team, We have one experiment running for our us...

wonderful-beach-39533

08/12/2026, 10:14 AM

Hi Team, We have one experiment running for our users. I was going through the experiment data and found some anomalies. There are some users who were supposed to be a part of the experiment as per the rules setup but we didn't receive any

experiment_view

callback stating user has entered the experiment. Is there any way growthbook team can help us with logs on their end for a couple of users to understand if they went into that experiment or not? Will help us understand the anomaly between GB experiment attribution and our internal tracking. Let me know what all information is required to debug this further. Happy to share.

flaky-noon-11399

08/12/2026, 1:05 PM

Hi @wonderful-beach-39533 GrowthBook doesn’t receive or store individual user attributes or assignment decisions, so we won’t have server-side logs showing whether a specific user entered the experiment. The evaluation happens inside your application, and the

trackingCallback

only fires when the feature is evaluated and the user is included. Common causes include the feature not being evaluated, a missing hash attribute, an earlier rule taking precedence, attributes being set too late, or the callback firing but the analytics event being dropped.

flaky-noon-11399

08/12/2026, 1:05 PM

To narrow this down: 1. Add a temporary log immediately before the feature evaluation showing the user’s attributes, especially the hash and targeting attributes. 2. Add a log as the first line of

trackingCallback

, before calling your analytics provider. 3. Check the evaluation result using your SDK’s debug output. For web SDKs, enable Dev Mode and use the GrowthBook DevTools extension. 4. Confirm the experiment rule is enabled for the environment, the hash attribute is populated, the user is within coverage and no earlier feature rule takes precedence. 5. Reproduce with the same user attributes and feature payload. If the SDK shows the user was excluded, the evaluation result should identify which condition failed. If they were included and the local callback log exists, but

experiment_viewed

is missing, the issue is downstream in the analytics pipeline.

10 Views

Open in Slack

Previous Next