Define the Few Metrics Your Funder Wants

How many metrics do you actually need?

Fewer than you think — usually one number for each claim your funder has to believe. The skill here is subtraction. A report with six well-chosen, credible numbers beats one with sixty that nobody can trace.

In the last step you wrote the report backward and ended with a short list of claims: who you reached, what changed, and the proof behind each. Now you turn each claim into a single metric you’ll actually measure — and leave everything else off the list.

A good metric names who, what changes, and by when

Vague metrics (“improved wellbeing”) can’t be proven. A usable one says who is measured, what specifically changes, and over what period — for example, “the share of participants who move into stable employment within six months of finishing.” That precision is what lets a number be checked later instead of argued about.

It’s also worth separating what you did from what changed. Funders have learned to discount activity counts (“500 people trained”) and reward outcomes (“of those trained, this many kept the job at six months”). Keep a couple of activity numbers for context, but let the outcomes carry the story.

Try it: turn your claims into a metric list

Paste the report outline from the previous step. This prompt proposes one metric per claim and flags which are real outcomes.

From the report outline below, propose exactly one metric for each claim. For each metric, write it as: who is measured · what changes · by when. Mark each as OUTPUT (something we did) or OUTCOME (something that changed for a person). If a claim can’t be measured yet, say so plainly rather than inventing a metric. Return a short table: Claim · Metric · Output/Outcome. Outline: [paste from step 1]

You’ll get a tight, sensible list — a good thing to bring to your team.

Where a chat tool stops

A general model will happily propose metrics, but it can’t hold them still. Ask it twice and the wording drifts — “employment rate” becomes “job placement rate” becomes “% employed.” It will also invent targets you never set. And because it doesn’t know your actual survey fields, none of these metrics connect to anything you collect. You end up with a nice list in a document that no system enforces.

That gap — a metric that means one fixed thing, everywhere, every time — is the whole point of the next step.

How the metric becomes real

In Sopact Sense, each metric on your list becomes a defined field with one meaning. The same “stable employment” is counted the same way in every program, every cohort, and every year — so your numbers are comparable instead of merely similar. As data arrives, those metrics are tracked automatically, not recomputed by hand each reporting cycle.

When this helps — and when it doesn’t

Do this whenever more than one person, program, or year will report against the same idea — that’s when loose definitions quietly break comparability. For a one-time, single-program count, a short list in your head is fine.

Next: Give Every Number One Definition → · or see how Sopact Sense tracks your metrics →