Change.org's email program reached 20 million highly engaged members across dozens of issue areas, from environmental policy to criminal justice reform to local community campaigns. I was the email marketer on a four-person team responsible for the entire program.

The core challenge was relevance at scale. Twenty million people don't care about the same things. Blasting the full list was a fast track to unsubscribes. But with members spread across dozens of issue verticals, each with different levels of passion and engagement, the targeting problem was genuinely complex.

Machine learning meets editorial strategy. We built a system that combined machine learning with hands-on editorial judgment to match content to members by issue affinity, engagement behavior, and lifecycle stage. The ML models handled pattern recognition at a scale no human team of four could manage. The editorial layer ensured that what we sent actually made sense contextually, not just statistically.

Making every send count. At 20 million members, even marginal improvements in targeting had outsized impact. A small lift in open rates meant hundreds of thousands more people engaging with petitions. A reduction in fatigue-driven unsubscribes meant the list stayed healthier over time. The discipline was constant: every email needed a clear audience, a clear purpose, and a reason to exist in someone's inbox that day.

Supporting the broader ecosystem. The email program didn't operate in isolation. It was the primary engagement channel feeding petition signatures, donations, and deeper platform engagement. The work required close coordination with product, campaigns, and data teams to ensure email strategy aligned with what was happening on the platform and in the world.

The result was an email operation that kept 20 million members engaged across a wildly diverse set of issues without burning the list down.