Every funder eventually has to answer an uncomfortable question: how do you know your money did anything?
The easy answer is to count what's countable. Dollars granted. Number of organizations funded. Attendees at a program. Those numbers are real, and they're not meaningless, but they're also the easiest ones to report because they say almost nothing about whether anything actually changed for anyone.
I learned this the hard way, sitting across from community organizations trying to fill out a reporting template that asked them to reduce years of relationship building and trust repair into a single output number. The template wasn't wrong to want accountability. It was wrong about what accountability should look like.
The metrics a funder chooses are never neutral. They're a statement of values, whether anyone intends them to be or not. If you only measure how fast money moves out the door, you're telling your team that speed matters more than fit. If you only measure the number of grantees, you're rewarding breadth over depth, even when depth is what a community actually needs. Every KPI is a small argument about what counts as success, and most organizations pick theirs without ever having that argument out loud.
Working on Edmonton's Vital Signs research taught me to sit with a harder version of this question: not just what should we measure, but who gets to decide what "better" even means for a given community. The data we pulled together, on housing, on belonging, on economic security, told a clear story on paper. But the story only became useful once we brought it back to the people living inside it and asked whether it matched what they actually experienced. Sometimes it did. Sometimes the numbers were technically accurate and still missed the point entirely.
That's the tension at the center of impact measurement. You need numbers, because without them you can't tell a good program from a well marketed one, and you can't defend your budget to a board or a government partner. But numbers alone will always flatten something. The task isn't to choose between rigor and nuance. It's to build a measurement practice that holds both: hard data, and the harder, slower work of actually listening to the people the data is supposed to represent.
A few things I've come to believe on this, none of them original to me, all of them earned the slow way.
Measure outcomes, not just outputs. Outputs tell you what you did. Outcomes tell you whether it mattered. They're not the same question, and reporting one instead of the other is how organizations end up technically successful and practically useless.
Ask the community what they'd measure before you hand them your template. You'll usually learn that the thing you thought was the win isn't the thing they experienced as the win.
Build in room for the metric to be wrong. If your KPIs never surprise you, you're probably not measuring anything real. The moment a number contradicts your assumptions is the moment it becomes useful.
Report what you got wrong, not just what you got right. Funders and organizations that only ever show success are, by definition, hiding their failures, and failure is where the actual learning lives.
None of this makes measurement easy. It just makes it honest, and honest is the only version of it worth doing.