Product Analytics for Growth Teams
PostHog instrumentation, warehouse modeling, dashboards, experimentation, and an AI-powered query agent — one analytics hub that covers the full lifecycle from event to insight.
The Analytics Stack
PostHog for event tracking and product analytics. BigQuery as the warehouse. dbt for data modeling. Stripe joined to product events for revenue analytics. AI agents for natural-language querying.
Core Analytics
- Trends and active users: DAU/WAU/MAU, signups, sessions, purchases over 7/30/90 days. Stickiness and returning users.
- Conversion funnels: Step-by-step drop-off from landing to signup to activation to revenue. Break by country, device, source, cohort.
- Activation and onboarding: Define and measure your activated user. Find where new users quit.
- Retention and lifecycle: D1/D7/D30 retention by cohort. Segment users as new, returning, resurrected, dormant, or churned.
- Cohorts and behavioral discovery: Compare power users vs new users vs payers. Find users who did X but not Y.
- Natural-language questions: Ask in plain English. The engine generates the query, runs it, and explains the answer.
Dashboards
- Product health: DAU, sessions, activation, retention, errors, revenue — one screen for Monday reviews.
- Monetization: Purchases, ARPPU, payer conversion, revenue by cohort and channel.
- AI cost: Cost by model, feature, user, prompt, or flow. Budget vs spend, week over week.
- Investor update: MAU, WAU, retention curves, growth rate, revenue, % activated. Auto-pulled monthly.
Experimentation
- A/B experiment results: Conversion, exposure, and metric results with clean readouts.
- Funnel and retention experiments: Test whether a change improves the funnel or D1/D7 retention.
- Revenue experiments: Compare purchase conversion or ARPU by variant, with proper exposure accounting.
- AI prompt experiments: Compare prompts and models by quality, cost, latency, and downstream conversion.
Web Analytics
- Traffic and sources: Sessions, visitors, pageviews. Source/medium/campaign performance.
- Landing page performance: Which pages attract and which convert.
- Web to app conversion: Landing page to signup to activation to revenue, closed loop.
Engagement Tiers
- Diagnostic (2 weeks): Audit the event taxonomy, dashboards, and experiment hygiene. Report on what to trust, what to fix, and the three biggest leaks.
- Build (12 weeks, 10 hrs/week): Embedded as fractional AI product lead service. Event spine + core dashboards + experimentation live. NL query agent shipped.
- Operate (monthly): Monthly readout on metrics, experiments, and what to test next.
Frequently Asked Questions
What does the product analytics hub include?
PostHog instrumentation, BigQuery warehouse modeling with dbt, core dashboards, A/B experimentation with automated readouts, web analytics, and a natural-language AI query agent.
What analytics stack is used?
PostHog, BigQuery, dbt, Stripe (joined to product events), and AI agents (Claude/GPT) for natural-language querying.
What types of experiments can you run?
A/B tests on funnels, retention, and revenue. AI prompt experiments comparing prompts and models by quality, cost, latency, and conversion.
Get in touch | Back to Home