Why are my metrics lower in GA4 than in GrowthBook...
# ask-questions
Why are my metrics lower in GA4 than in GrowthBook? (@tall-branch-42668)
Different User Counting Methods: GrowthBook might be counting users differently than GA4. For example, GrowthBook could be including users that GA4 filters out due to various reasons such as bot filtering or data processing differences.
Date Range Discrepancies: If the date ranges being compared in GA4 and GrowthBook are not the same, this could lead to differences in the metrics reported. It's important to ensure that the same date range is being used in both platforms for accurate comparison.
SDK Implementation Changes: If there have been recent changes to the SDK implementation that have not been reflected in GA4, this could cause discrepancies.
Ad Blockers: While ad blockers typically would cause metrics to be lower in GrowthBook, if GA4 is more affected by ad blockers than the method used by GrowthBook to track users, this could result in lower metrics in GA4.
Data Synchronization Issues: There could be issues with how data is being synchronized between GA4 and GrowthBook. For example, if GrowthBook is processing data more/less frequently or differently than GA4, this could lead to discrepancies.
Experiment Configuration: The configuration of the experiment itself might be different between GA4 and GrowthBook. For instance, if the experiment targeting or segmentation is not consistent across both platforms, this could result in different user counts.
Metric Conversion Window: The metric conversion window set in GrowthBook might be different from the one in GA4, leading to a different set of conversions being counted in each platform.
Data Source Configuration: Incorrect configuration in the data source connection between GA4 and GrowthBook could lead to discrepancies. For example, make sure the default dataset is the correct one.
Another possible explanation is the sampling with GA4. If the number of users exposed gets high, GA4 will use hyperLogLog to do some sampling and not actually give you numbers. It is possible that this is the cause of your user counts mismatches.
Thank you all! Date ranges have been accounted for We are not using hyperLogLog/other sampling functions to reduce the number shown in GA We are querying the same dataset in GB and in GA Not using targeting or segmentation other than user_pseudo_id By definition we exclude bots/web-crawlers from the Metrics tracked since the main Metric in focus is trial_starts, bots don't start those because a trial_start requires an account and payment info The adblockers is a good point (altho definitely doesn't make up for a ~70% mismatch) and one we were discussing - GA and
can be blocked/not initialize from the presence of these. We were wondering if GB also doesn't initialize from common adblockers? Or have you not seen that yet? Additionally I had a question around controlling for users on different devices: if a user sees 2 variants on 2 different devices due to not being logged in on one of them, once they log in does GB have any kind of merging recognition? We haven't really broached that topic yet before but would be good to have some background on what's possible from a GB side to account for dupes like that
we haven't yet seen GrowthBook be blocked
but again, the data is sent through GA, so I'm not sure how it could be so much larger
Yeah we are also going to run the same GB query in GA and see if we get the same result, since we're querying the same dataset