Hey all 👋 looking for some guidance on a segment + snowflake best practices. Segment recommends setting up QA and Production apps as different sources, and this results in one schema per application+environment combination when the data ends up in Snowflake. Growthbook appears to require the schema info to connect to a data source, so we ended up with 4 total data sources (Prod/QA and Frontend/Backend split). We've found that this doesn't play very well with features, for example - creating a/b experiment overrides end up tied to a single Datasource for prod and qa.
Is there a better way to think about organizing our data? We've also considered creating views to consolidate prod and qa schemas into single ones, though I don't believe this solves the issue about experiment override creation. Perhaps views by application are better?
04/06/2022, 8:54 PM
As of today, multiple data sources is the only real solution. We're actively working on a better way to do this by allowing a single data source to have multiple experiment queries and you can choose on an experiment-by-experiment basis which one to point to
04/06/2022, 8:59 PM
Right on - thanks Jeremy. Is the split we have setup now the best option at the moment, in your opinion? Have you seen others approach the division any differently?
04/06/2022, 9:10 PM
Usually experiment data from dev/staging is not that useful beyond just debugging that the tracking is working.
There will only be a few users and the behavior isn't representative of production, so the metrics and stats are not going to mean much