Learning Product Analytics Tips

Q: I'm learning product analytics. Any tips?
‍

Avoid the mistakes I made:

1. Poor Instrumentation

From conversion tracking to events, triple check that they are setup correctly. Employ 2 tools to cross-match results if you have to.

Nothing is worse than discovering the data you've collected for ages was inaccurate and polluted.

2. Lack of playbooks

Measurement is not the goal. A tidy dashboard with 10 dials is useless until it informs action.

Have a playbook of next steps for every anomaly you detect.

Traffic tanked? Conversions flatlined? Feature adoption zeroed out?

Line up the next actions:
- What question will you ask (e.g. did the site have downtime? )
- Who you'll ask (e.g. a tool or team member)
- Counter-measure (e.g. change the copy, call in a meeting etc.)

3. Lack of context

Simply put, all your time series & analytics should be annotated with the dates releases went out, external events happened, marketing activity etc.

Numbers without context is like watching a 3D movie without the goggles. You have an idea of what's going on but the picture is blurry.

4. No segmentation

Aggregation masks the underlying truth. Insightful trends usually hang out with segments & cohorts: demographic, business-specific and time-based.