I have worked on both ends of public decisions: in the rooms where they are shaped, and in the neighborhoods where they land. The most important lesson I took from that is simple. Policy is a promise. It is kept or broken in the last mile.
Where the last mile breaks
When I worked on the Affordable Care Act rollout at EmblemHealth, the law had already been passed and the rules written. What remained was the part that decided whether any of it mattered: whether people understood what was available, trusted the messenger and could actually complete the process.
Inside the Department of Correction, advising leadership on Rikers Island reform, I saw the same thing from another angle. A reform could be sound on paper and still stall on staffing, on operations, on incentives no one had mapped, or on the gap between how a policy was described and how it was experienced.
In both cases the design was not what failed. The failures came in the last mile:
- Translation. The people affected never heard the policy in terms that made sense to them.
- Trust. The messenger lacked credibility where it mattered.
- Capacity. The front-line institutions, public or community-based, didn't have what they needed to deliver.
- Feedback. Nobody measured what happened at the point of contact, so nobody corrected course.
Community institutions are last-mile infrastructure
Governments often treat community organizations as vendors: useful for outreach, easy to replace. That is a strategic mistake. Trusted community institutions are last-mile infrastructure. They have the relationships, the credibility and the ground-level knowledge that decide whether public investment turns into public outcomes.
That means investing in them as infrastructure: multi-year partnerships instead of one-off contracts, shared outcome measures instead of activity reports, and a real role in designing programs, not just delivering them.
A practical test
Before any major initiative launches, I would ask three questions:
- Who will explain this to the person it is meant to help, and do they trust that messenger?
- Which institution on the ground will carry it, and what does it need to succeed?
- How will we know, within months rather than years, whether it is working where it lands?
If those answers are vague, the policy isn't finished yet, however strong the design.