Subscription Funnel Analysis (Practice Project)
A self-directed SQL and Python exercise — analyzing a public funnel dataset to practice finding and explaining a real drop-off point.
- Role
- Practice Project
- Year
- 2026
- Tools
- SQL, Python, pandas, Statistics
Why This Project
I wanted practice with the core analyst workflow — going from raw event-level data to a specific, defensible recommendation — using a public dataset rather than invented numbers. This is a learning project, not a client engagement.
Approach
I queried event-level signup and onboarding data, built a stage-by-stage conversion funnel segmented by device type and channel, and used basic statistical testing to check whether the differences I was seeing between segments were meaningful or just noise.
What I Found
- Mobile users dropped off at onboarding at a noticeably higher rate than desktop users in this dataset — a pattern that pointed toward a UX friction issue rather than an acquisition-quality one
- Users who reached an early “activation” milestone quickly converted at a meaningfully higher rate regardless of channel
Reflection
The most useful part of this project for me wasn’t the SQL — it was practicing the discipline of checking whether a difference is actually significant before writing a recommendation around it. That’s a habit I want to bring into real analyst work, not just something I did once for a portfolio piece.
Links
Repo and write-up link go here once published.