Who this is for
This lesson is for product analysts and adjacent roles who use BI tools to track product health, explain changes, and help teams act. It also fits PMs who want a reliable view of activation, retention, and revenue.
Prerequisites
- Comfort with a BI tool (Looker, Tableau, Power BI, Metabase): filters, time controls, calculated fields.
- Basic understanding of events and users, or accounts.
- Familiarity with date logic: weekly and monthly aggregates, cohort by signup date.
- Optional: basic SQL or semantic modeling concepts (dimensions, measures, grain).
Why this matters
In the role, you will be asked to:
- Report yesterday's activation drop and show which segments drove it.
- Explain a retention change after a new onboarding flow.
- Track a feature adoption target and alert the team when trend breaks.
- Show weekly DAU, WAU, MAU, but also why they moved.
- Build a single source of truth that PMs and leaders trust.
Real task example: Activation down 6% week over week
What you do: check signup-to-activation funnel, segment by acquisition channel and device, compare last 2 weeks, and annotate the dashboard with the cause (for example, iOS crash) and action.
Concept explained simply
A product metrics dashboard is a clear, reliable board that shows what matters now, why it moved, and where to look next.
Mental model
- North star plus drivers: one guiding outcome at the top; below it, leading indicators and diagnostic views.
- Cube of funnel x lifecycle x segment: any metric should be sliceable by where users are in the funnel, where they are in lifecycle, and by key segments like channel or platform.
- Heartbeat: a cadence to refresh and review (daily or weekly) with consistent definitions and time grains.
Core metrics and definitions
- Signups: number of new accounts or users created in a period. Choose user or account as your primary entity and be consistent.
- Activation rate: activated users divided by new signups in the same cohort window (for example, activation within 7 days of signup). Define the activation event clearly (for example, completed key action X).
- DAU, WAU, MAU: distinct active users per day, week, month. Use rolling windows for WAU and MAU or calendar buckets; label clearly.
- Retention: share of a cohort that returns in later periods. For example, day 7 retention equals users in cohort who had any active event on day 7 divided by users in cohort.
- Conversion funnel: progression from stage A to B to C; conversion rate equals count at next stage divided by count at previous stage, for the same cohort.
- Churn (customer): customers lost in period divided by customers at start of period. Churn (revenue) is revenue lost divided by starting revenue.
- ARPU or ARPA: average revenue per user or account; total revenue divided by count of active users or accounts.
- Segments: channel, device, geo, plan, cohort month. Avoid too many by default; expose as filters.
Lagging vs leading metrics
Lagging: revenue, MAU, churn. Leading: activation rate, setup completion, feature adoption. Build the top with lagging metrics and a second band with the leading drivers.
Dashboard structure and patterns
- Top band: north star and 3 to 5 key KPIs with sparklines and week over week deltas.
- Funnel: signup to activation to engaged to paid; show rates and counts; allow cohorting by signup date.
- Retention: cohort heatmap (rows cohort week, columns week since signup, cells percent retained).
- Segments: a ranked bar chart of top segments contributing to change (for example, channels by activation shortfall).
- Annotations: space to write what changed and the owner.
- Controls: date range, time grain, entity (user vs account), segment filters.
- Governance: metric definitions, owner, refresh cadence, last updated timestamp.
Visual tips
- Use consistent date grains across tiles. If top KPIs are weekly, keep diagnostic views weekly or clearly labeled.
- Prefer percent and count side by side for funnels to prevent misreads.
- Heatmaps for cohorts, funnels for stages, lines for trends, bars for segment comparison.
- Show confidence bands only when you have a valid sampling method; otherwise remove to avoid false precision.
Worked examples
1) Signup to activation funnel (weekly)
Goal: see if onboarding improved activation.
- Define cohort: users who signed up in the selected week.
- Define activation: user completed key action X within 7 days of signup.
- Create a funnel: signups → activated; show counts and rate.
- Add segment picker: acquisition channel.
- Compare last 2 weeks with a simple table of rate delta by channel.
Interpretation: If overall activation improved but one channel dropped, focus onboarding changes or technical issues for that channel.
2) Retention cohort heatmap
Goal: track new user week 1 to week 8 retention.
- Rows: cohort week (signup week).
- Columns: week number since signup.
- Cell: percent of users active in that week.
- Tooltip: count active and total users.
- Segment filter: device type.
Interpretation: A diagonal pattern improving after a release suggests better onboarding or sticky features.
3) Revenue and churn panel
Goal: link user metrics to money.
- MRR trend with week over week delta.
- New MRR vs churned MRR bars; net = new minus churned.
- ARPA trend by plan.
- Churn rate table by cohort month and plan.
Interpretation: A stable activation but rising revenue churn may point to billing or value gaps for a specific plan.
Build it: step by step (tool agnostic)
- Pick entities and grain: decide user or account as the primary entity and use weekly grain for the core board.
- Define metrics: write plain-language definitions with formulas (for example, activation within 7 days of signup).
- Prepare fields: date keys, cohort labels, event flags (did_activate = 1 or 0).
- Create top KPIs: DAU or WAU, activation rate, week 1 retention, MRR.
- Add diagnostic tiles: funnel, cohorts, segments.
- Add controls: date range, segment filters, time grain switch if needed.
- QA: sample 5 users across cohorts and confirm activation and retention flags match raw events.
- Annotate and ship: include owner, definitions, and last updated time.
Data prep guide
Minimum columns helpful across tools: user_id or account_id, signup_date, event_date, event_name, channel, device, plan, revenue_amount. Derived: cohort_week, is_active_week, activated_within_7d, week_number_since_signup.
Exercises
These mirror the tasks below in the Exercises section of this page. Do them inside your BI tool or a spreadsheet if needed.
- Exercise 1: Build a minimum product health dashboard with 5 KPIs and 2 diagnostics.
- Exercise 2: Add segmentation and thresholds; annotate insights.
- [ ] Keep entity consistent (user or account) across tiles.
- [ ] Label time grain on every tile.
- [ ] Show both counts and rates where applicable.
- [ ] Provide filters for the 2 or 3 most important segments.
- [ ] Add last updated and owner.
Ready-to-ship checklist
- [ ] North star and 3 to 5 KPIs at the top with deltas.
- [ ] Funnel and retention included.
- [ ] Segments visible but not overwhelming (exposed as filters).
- [ ] Clear metric definitions and owner in a definitions panel.
- [ ] QA notes: sampled users match flags and cohort labels.
Common mistakes and self-check
- Mistake: mixing user and account denominators. Fix: choose one entity and label it on the page.
- Mistake: activation calculated across calendar weeks instead of within user cohort. Fix: define activation within X days of signup and compute per cohort.
- Mistake: retention misread due to missing zeros. Fix: include zero-value weeks and consistent cohorts.
- Mistake: stale data without notice. Fix: show last updated and expected refresh cadence.
- Mistake: too many default segments. Fix: keep top 3 as filters and use a detail view for more.
- Mistake: time grain mismatch. Fix: align all tiles to weekly or label exceptions in the title.
Self-check prompts
- Pick 3 random users from last week. Do their activation flags match the events?
- Change the date range. Do rates stay consistent with counts and denominators?
- Does the funnel rate equal the ratio of the counts shown for the same cohort and window?
Practical projects
- Weekly product health board: WAU, activation, week 1 retention, funnel, segment breakdown.
- Activation deep dive: test alternative activation definitions and pick the most predictive.
- Cohort retention explorer: add feature adoption tags and compare cohorts before and after a release.
- Feature adoption scoreboard: track adoption by plan and device with target thresholds.
- North star drivers tree: link the north star to measurable leading indicators.
Learning path
- Define the north star and core KPIs with precise formulas.
- Build the Minimum Viable Dashboard: top KPIs, funnel, retention.
- Add segments and diagnostics: contribution analysis and cohort heatmap.
- Instrument governance: owner, definitions, refresh cadence, annotations.
- Automate and alert: thresholds for activation and retention breaks.
- Advance: add experiment overlays and revenue bridges.
Quick test and progress
The quick test below is available to everyone. Log in to save your progress and resume later.
Mini challenge
Your PM believes iOS activation is the main problem. Using the dashboard you built, write one sentence with the current activation rate and the top 1 driver of the change (segment and delta). Add a second sentence with your proposed action and owner.
Example answer style
Activation is 42 percent this week, down 4 points week over week, driven by iOS at -9 points from channel X. Action: investigate onboarding crash; owner: Mobile lead.
Next steps
- Turn on a weekly review ritual: snapshot KPIs, add one insight and one action each week.
- Add alert thresholds for activation and retention changes beyond agreed limits.
- Document metric definitions in the dashboard so new teammates can self-serve.