Denys

2024–2025

Analytics-driven product decisions with PostHog

Set up product analytics from zero and used it to kill, reshape and prioritise features.

Role
Design Lead — analytics ownership, event taxonomy, dashboards, research synthesis
Team
1 PM, 1 engineer for instrumentation
Timeline
Ongoing since [month] 2024

Context

[ The state before: no product analytics, decisions made on the loudest customer or founder intuition. Why you, the designer, picked this up. ]

Problem

[ The concrete cost of not knowing. A feature built on a guess that nobody used; a debate that ran for weeks because nobody had numbers. ]

My role

[ What you owned: choosing PostHog, defining the event taxonomy and naming convention, writing the tracking plan, building dashboards, running session-replay reviews, and turning findings into decisions the team acted on. ]

Process & key decisions

The tracking plan

[ How you decided what to track (jobs, not clicks). The naming convention. How you kept engineers from adding random events. ]

Event taxonomy
The event taxonomy — object_action naming, grouped by job.

Decision changed 1 — [feature you killed or paused]

[ What the data showed, what the team believed before, what you did. ]

Decision changed 2 — [feature you reshaped]

[ A funnel or replay finding that changed a design. ]

Decision changed 3 — [priority you moved]

[ Where analytics changed the roadmap order. ]

Activation funnel dashboard
The activation funnel the whole team watches — [caption].

Solution

[ The rituals that stuck: weekly metric review, replay Fridays, the dashboard every PM opens. What makes it a practice rather than a tool. ]

Impact

3
Roadmap decisions reversed by data
In the first two quarters
+14%
Activation rate
After the funnel-led fixes
< 1 day
Time to answer a 'do users…?' question
Was: never

[ Replace with real figures. ]

What I'd do differently

[ E.g. instrument earlier; involve engineering in taxonomy sooner; fewer dashboards. ]