Bernard Marr has published more KPI suggestions than there are species of beetle. At some point, someone presumably looked at the list and thought: we need a bigger boat.
The genre is well established. Here are 75 KPIs for retail. Here are 50 for healthcare. Here are 127 for whatever your organisation does this quarter. Pick the ones that sound right, plug them into your dashboard, and performance management is solved. Marr is not alone in this — he is simply the most prolific example of an entire publishing and consulting category built on the same premise: that the measurement problem is a selection problem.
It isn’t.
Now AI has entered the room. Naturally, it has been applied to the selection problem with great enthusiasm. Feed a language model your strategic priorities and it will return KPIs — faster, more numerous, and more confidently phrased than any lookup book. Some tools will even benchmark them against industry norms.
What none of them will tell you is whether the measurement counts as evidence.
That is not a criticism of the technology. It is a description of the problem the technology inherited. AI didn’t create the epistemic gap at the centre of performance measurement — it just gave it a faster engine.
Here is the gap.
Seventy years of performance measurement literature produced tools for evaluation. Scorecards, OKRs, KPI frameworks — all of them answer a version of the same question: how are we doing? They are designed to render a verdict.
Evidence instruments answer a different question: what actually happened, and can we show it? They are designed to produce something scrutinisable — a record that holds up when someone who wasn’t in the room asks to see the basis for a decision.
The difference sounds subtle. In practice it is the difference between a governance document and a slide deck.
W. Edwards Deming understood this distinction. He understood that measurement without statistical grounding produces noise dressed as signal — that what gets reported as performance data is often variation that was going to happen anyway, attributed to decisions that had nothing to do with it. The field read Deming, built better dashboards, and moved on. The question he was actually raising got filed under “manufacturing” and largely left there.
That question was: what does it take for a measurement to count as evidence?
Marr’s books don’t answer it. They can’t — it isn’t a question that has a lookup answer. It requires a framework with a defined cascade, pre-declared thresholds, and calibration logic that’s logged before deployment, not reverse-engineered after the results come in.
It requires, in other words, exactly the kind of structural discipline that makes lookup KPIs feel like too much work — until the moment you need to defend a resource decision to a board, a donor, or a regulator, and the dashboard can’t tell you anything that wasn’t already assumed.
AI accelerates whichever approach you bring to it. If you bring it a well-designed evidence architecture, it can do useful things inside that structure. If you bring it a list of KPIs that sounded right in a meeting, it will help you produce more of them, faster, in a font of your choosing.
The MetriqOne framework was built from the question Deming raised and the field largely dropped. Not because lookup books are useless, but because selection was never the hard part. The hard part is calibration — knowing in advance what your indicators are supposed to detect, under what conditions, and what it means if they don’t.
That answer is in Foundations of Evidence-Based Leadership (Book II), which traces the epistemic failure across five major measurement traditions and locates exactly where the field went wrong.
The deployment answer — how to build a cascade that produces scrutinisable evidence in organisations without stable infrastructure, clean data, or specialist software — is in Proof Without Judgement (Book I).
Both are available on Amazon. Neither contains a list of 75 KPIs. That is a feature, not an oversight.
Torgeir Skogvold is the founder of MetriqOne and author of the MetriqOne Trilogy: Measurement for the Real World.