We are very good at starting things. We are much worse at letting good ones grow.
Walk into almost any agency and the evidence is there: the dashboard that worked beautifully in one district, the cash model that cut costs in one response, the local partnership that every evaluation praised. Then the project closes. The findings get filed. Two years later, someone designs the same idea from scratch, somewhere else, as if the first one never happened. We call this learning. It behaves more like amnesia.
This is not a shortage of good ideas. It is a funding structure that rewards starting and offers almost nothing for scaling.
Money tends to arrive in short cycles tied to single grants, so an innovation lives or dies with one budget line. The incentive that structure creates is to show novelty in a report, not durability in a budget. A pilot that works earns no automatic claim on next year’s core funding, so even a proven model can have nowhere to go. And the cost of standing still stays invisible. Nobody is asked to account for the reach and the savings we walk away from when a working idea is quietly shelved. We count the launch and we never count the abandonment, so the launch becomes the only thing the system reliably produces.
Scaling is not a bigger pilot. It is a different set of conditions, and most of them are about how we structure money, evidence, and ownership, not about how clever the idea is.
Fund the transition, not just the test. A pilot budget should carry a defined path to either core funding or a clear sunset, decided on stated criteria. That way success has somewhere to go and an ending is named honestly instead of left to drift.
Count the cost of not scaling. If we can estimate the reach and the savings a proven model would deliver at full size, we can set that figure beside any decision to drop it. What gets counted gets defended.
Give the idea an owner beyond the project. An innovation attached only to a grant ends with the grant. One held by a team, a standard, or a shared system can outlive any single funding cycle.
Lower the cost of adoption for the next user. Most of us do not need to invent. We need to borrow well. That means publishing not only what worked, but the unit costs, the failure points, and the conditions a model needs, so the second team can start from the second mile instead of the first.
None of this requires more money. It requires treating scale as a real stage of work, with its own funding, its own evidence, and its own owner, rather than as something we hope happens on its own once the report is filed.
So here is the test worth holding ourselves to. A pilot that cannot describe how it graduates is not yet an innovation. It is a rehearsal. And the people we serve deserve better than a sector that rehearses forever.