Written by: David Carneal – Digital Efficiency Consulting Group – DECG
Read Time: 2 min
If your KPI triggers a debate about what it means, you do not have a KPI. You have a mood ring.
Dashboards fail when the metric cannot defend itself. "Proof metrics" is the idea that a KPI should come with enough context that a smart person can trust it without guessing.
The 4 proof questions (non-negotiable)
1) Definition: What exactly counts and what does not?
Example: "On-time delivery" sounds obvious until you learn half the company means "ship date" and the other half means "customer receipt date."
2) Source: Which system is the authority?
If one report pulls from the ERP and another pulls from a spreadsheet that someone updates on Tuesdays, you are not measuring, you are improvising.
3) Cadence: How often does it refresh?
A monthly report does not help you run a weekly process. A daily refresh might be noise if the underlying work moves weekly. Cadence must match decisions.
4) Owner: Who is responsible for the metric being accurate?
Ownership is not blame. It is stewardship. Someone should know how the number is made.
The "proof pack" you can attach to any KPI
Put this in a one-page note (or a short internal wiki page):
– KPI name and plain-English definition
– Calculation (formula, fields, filters)
– Source system(s) and report link
– Refresh cadence and timing
– Known caveats (lags, exclusions, seasonality)
– Metric owner (name or role)
The payoff: fewer fire drills
When a metric has proof, your meetings change:
– Less time arguing about the number
– More time discussing what to do about it
– Faster root cause discovery because inputs are visible
A small warning
Proof does not mean perfection. It means the KPI can be trusted enough to make decisions, and errors get found quickly instead of lingering for months.
CTA: Audit one KPI: can you name the source system, definition, refresh cadence, and who owns it? If any of those are fuzzy, the KPI is not ready for decision-making yet.