I’ve been pushing internally at our company to use Metabase. Adoption from technical people is rather good, the UI makes sense to them and they understand the underlying data model. For less technical people, one consistent feedback was that it was impossible to do funnel analysis in Metabase.
They compared the experience to tools like Amplitude, where it’s rather easy to do it: you select a segment of users, you choose the events your want to follow in the funnel, you have a bunch of parameters to configure the funnel. But it makes sense to a non-technical human. So I decided to show them the way and build a funnel in Metabase. Honestly, it’s super hard, even for a technical person (I’ve 8+ years experience as a software engineer; been using Metabase for 1.5 years) and really depends on the data model (we funnel data from Segment to a Postgres database; Metabase sits on top of this database). Here is how the query looks like in the editor:
- 5 left joins on the User id key
- Filter: a complex constraint to make sure that the user did the event in the right order (and of course, you could have any variation of this)
- Summarize: having to distinct values of User ID is not super intuitive when we want counts. And we need to group by something so that we can display it on a line chart (here context library name is the same value for every row)
- writing this in SQL would be easier than with the editor from my POV
So is there a plan to improve this experience someday? I didn’t see anything in your roadmap on GitHub that talks about funnels and the main topics about funnels in the forum are about formatting the data for the funnel graph type.