Hello, I'm new to experimentation. We are trying to setup a data pipeline and integrate it with Growthbook. I'm trying to understand the best way to model the warehouse.

Assignments Table
- user_id
- timestamp
- date(partition)
- hour (partition)
- experiment_id (partition)
- variation_id
- browser
- country

Events Table
- user_id
- timestamp
- date (partition)
- hour (partition)
- value

Dimension Table
- user_id
- value

Questions: 1. Do we create per metric events table ? For example, Revenue Events Table, Impressions Events Table 2. Is it okay to have one big assignments table ? 3. I'm assuming we have many dimension tables of the format mentioned above ? could you please provide guidance on this as I'm doing this for the first time ?

1. This depends a bit on your needs - but typically if you use an event tracking system like Segment, Rudderstack or Jitsu, they’ll make a table per event type. You could also throw this into one event table, but depends on the volume of traffic. 2. Yes, it’s unlikely to get super large. 3. Dimensions are okay as a separate table, but can also be attributes of the other tables - like in a user table (assuming you have one), or even in the assignment table (or both).