Part 4 · 1 chapters · ~8 min
Reading Data and Talking to Users Yourself
Weekly habits: reading support tickets, watching session replays (with privacy), joining interviews, querying the data warehouse yourself, funnels and cohorts for your feature, and turning what you see into small fixes and bigger bets.
8
A weekly routine
| habit | time | what it gives you |
|---|---|---|
| read 20 support tickets for your area | 30 min | real problems in real words, including bugs no alert caught |
| watch 5 session replays (masked) | 30 min | where people hesitate, rage-tap, or give up |
| join one user interview a fortnight | 45 min | context: why, not just what |
| check your feature's funnel and a cohort | 20 min | whether usage is growing, flat or decaying |
| write three observations in the team channel | 10 min | shared learning, and small fixes that get picked up |
code
-- a funnel for the letter feature, last 28 days
SELECT count(DISTINCT user_id) FILTER (WHERE event = 'letter_entry_viewed') AS viewed,
count(DISTINCT user_id) FILTER (WHERE event = 'letter_eligibility_checked' AND (props->>'eligible')::bool) AS eligible,
count(DISTINCT user_id) FILTER (WHERE event = 'letter_generated') AS generated,
count(DISTINCT user_id) FILTER (WHERE event = 'letter_downloaded') AS downloaded
FROM events WHERE ts > now() - interval '28 days';