Senior Product Manager · 2022 — 2024
Amplitude CDP Integrations Factory
CDP ARR
$6M → $21M
Partner integrations shipped
50+
Time-to-integrate reduction
Placeholder metric
Context
Amplitude was entering the CDP market in a competitive landscape dominated by Segment and mParticle. The integrations ecosystem was a critical differentiator — customers needed to connect Amplitude to their existing data stack, and the more connectors we had, the stickier the platform became.
Problem
The integrations backlog was 80+ requests deep. Engineering was building each integration one at a time, every one bespoke. There was no shared schema, no partner self-serve path, and no way to scale. We were losing deals because the integration customers needed wasn't available.
Approach
- 1Audited the integration backlog by revenue potential, customer frequency, and engineering cost — identified the top 15 high-ROI connectors to prioritise.
- 2Designed a JSON Schema-based integration spec that standardised how data mappings, auth flows, and sync configurations were described — enabling reuse across integrations.
- 3Built a partner portal that let technology partners self-certify integrations using the SDK, reducing the engineering touch required per integration.
- 4Launched a tiered partner programme: Certified, Verified, and Community — with clear SLAs and co-marketing criteria at each tier.
What I shipped
- ✓Integration factory architecture: shared schema, auth library, and validation tooling used across 50+ connectors
- ✓Partner self-serve certification portal with documentation, sandbox environment, and testing checklist
- ✓Tiered partner programme with onboarding playbook and go-to-market support for Certified-tier partners
- ✓Integration quality metrics dashboard for ongoing monitoring of sync health and error rates
Impact
- ·CDP ARR grew from $6M to $21M over 18 months — integration coverage was cited in 70%+ of CDP expansion deals.
- ·Time from integration request to GA dropped significantly after the factory model launched.
- ·50+ integrations shipped in 18 months vs 12 in the prior 18 — without increasing the integrations team headcount.
Learnings
- Schema design is strategy. The JSON Schema decisions we made early constrained everything downstream — investing 6 weeks in schema design before writing a single integration saved months of rework.
- Partners are customers. Treating the self-serve portal like a user-facing product — with real dogfooding and feedback loops — was the unlock for partner adoption.
- Metrics need to lag revenue, not lead. We initially tracked integrations shipped; switching to integration-influenced ARR changed how the team prioritised the backlog entirely.
One key tradeoff
Decision
Build a standardised schema vs. support fully custom integration contracts per partner
Rationale
Custom contracts would have let us onboard a few large partners faster, but created a maintenance nightmare at scale. Standardisation sacrificed short-term speed for long-term leverage.
The tradeoff
We lost 2 potential Certified partners in the first 6 months who couldn't adapt to the schema. In hindsight, the schema was correct and those partners would have been high-maintenance.