I've increased the query cache ttl from 60 to 720 minutes on our data source so it would reduce the fetches from the biguqery. When I inspect the queries that reach bigquery I see growthbook selects a datetime from-to with the to changing with each manual update. For us the intraday data is not that important, so 1 update a day is the maximum we need. This can be a manual update since we don't check the statistics each day. Can we optimize this is a way and is the query cache relevant for us in this case?

Hi @lively-tiger-66465 Happy Friday! Have you disabled the automatic updates under Settings > General > Experiment settings > auto update frequency? If not, you can select "Never", so that your experiment data is manually updated on an ad hoc basis. You can keep the query cache TTL fairly long (like a few hours), so if you or teammates open the same experiment again and hit Update without changing the date range, GrowthBook can just reuse the cached results instead of re‑querying BigQuery.

Hi, yes I found that setting and disabled it 🙂 For the query cache, I also made that longer, but it seems that's not working. Each time I press Update on an experiment, I see it hits my bigquery, probably because it's a from-to query and the "to" is alway a timestamp now().