Intent
Keep measurement in its place as one input among several, so that what is countable does not quietly become the whole definition of what matters.
The Pattern
Metric fixation is the belief that what is measurable is what matters most. It begins reasonably — you cannot manage what you cannot see — and ends with a number standing in for the thing it was supposed to describe.
Three well-worn observations describe the failure, and our glossary keeps all three together. Goodhart’s Law: when a measure becomes a target, it ceases to be a good measure. The McNamara Fallacy: relying on narrow metrics while ignoring the complex human realities those numbers fail to capture. And metric fixation itself, the belief that quantification is the only real form of knowing.
The mechanism is substitution. A number is chosen because it correlates with something valuable. The number becomes the target. People optimise the number. The correlation breaks — and the number stays, now measuring the optimisation rather than the value.
What gets erased is everything harder to quantify: regulation, relational work, creativity, care, resilience, dignity. None of these produce a clean figure, so in a fixated system none of them officially happened.
For neurodivergent people this is acute, because the metrics on offer tend to measure pace, consistency and availability — the axes on which variation is largest and least voluntary. A spiky profile scores badly on a measure that rewards flatness, whatever the work was worth.
As one of our own principles has it: when you measure, include the measurer. The choice of what to count is itself a decision, made by someone, in a position, with interests.
Why It Matters
When a metric becomes the target, people may:
- do the countable work instead of the valuable work
- be assessed on output while the conditions producing it go unrecorded
- have care, mentoring and repair treated as though they did not occur
- be penalised for variation the measure was never designed to tolerate
- watch a number improve while the thing it stood for gets worse
- learn to produce the figure rather than the outcome
The measure keeps reporting success, which is why fixation is so hard to dislodge from inside.
Forces
- Institutions must account for themselves, and numbers travel further than judgement.
- What is easy to count is rarely what is most worth counting.
- A target creates an incentive to optimise it directly.
- Qualitative evidence is dismissible in a way a figure is not.
- Measurement confers an appearance of objectivity on a subjective choice.
- The work that sustains a system is usually the least visible in its data.
Problem
How do we use measurement without letting the measure replace the purpose, or erase the work that will not reduce to a number?
Solution
Say out loud what the metric is a proxy for, and keep saying it. A proxy that cannot name its referent has already become the referent.
Use several measures that pull in different directions, so no single one can be optimised without cost. Pair every quantitative measure with a qualitative one that would notice its failure. Review whether the correlation still holds, and retire measures that have become targets.
Record the conditions alongside the output, because a figure without its context is not evidence. This is the counterweight to Meritocracy Trap and works with Energy Accounting.
What This Looks Like
In physical spaces
- Do not use attendance as the measure of engagement.
- Stop counting hours present as hours worked.
- Notice the person who keeps the room functioning and find a way to say so.
- Do not rank spaces by utilisation alone.
- Never let a headcount decide whether a quiet space is justified.
In digital spaces
- Do not measure contribution by message volume.
- Keep response-time statistics out of performance judgements.
- Avoid dashboards that make the countable look like the complete.
- Report uncertainty alongside the number.
- Do not let a metric be visible to management and invisible to the people it describes.
In social and organizational practice
- Name the purpose each metric serves, in writing, next to the metric.
- Ask what this number would look like if someone were gaming it.
- Collect one form of evidence that the number cannot produce.
- Retire measures on a schedule rather than accumulating them.
- Do not tie consequences to a proxy you have not validated recently.
- Never treat the absence of data as the absence of the thing.
Implementation Notes
Take your most consequential metric and try to break it on paper.
Ask:
- What is this a proxy for, and when did we last check the link?
- How would someone improve this number without improving the thing?
- What valuable work produces no signal here at all?
- Who chose this measure, and what does it make them look like?
- What would we notice getting worse before this number did?
- If the number and the people disagree, which do we believe?
A good test: if you cannot state what the metric is a proxy for without using the metric’s own name, the substitution has already happened.
Signals of Misuse
This pattern is being violated when people hear:
- “If you can’t measure it, it doesn’t exist.”
- “The data says otherwise.”
- “We need something objective.”
- “Hit the number and we’ll talk about the rest.”
- “That’s not something we track.”
- “Anecdote isn’t evidence.”
The last one is a real principle doing false work: it dismisses the only evidence a bad metric cannot suppress.
Consequences
When this pattern is honored:
- numbers inform judgement instead of replacing it
- care and relational work become visible as work
- gaming becomes harder because no single measure carries the weight
- variation stops reading as underperformance
- measures are retired before they become targets
- the choice of what to count is open to argument
When it is ignored, an organisation optimises itself steadily toward the wrong thing while its reporting improves.
Anti-Patterns
- A single headline metric with consequences attached
- Targets set on proxies nobody has validated
- Dashboards presented as complete pictures
- Qualitative evidence excluded as unrigorous
- Metrics visible upward and hidden downward
- Counting presence instead of contribution
- Treating unmeasured work as work not done
Design Moves
Strategies for measuring without fixating.
See recipes:
Related Environments
Environments governed by what they count:
Related Experiences
Experiences shaped by being measured:
Patterns Below
Patterns Above
The meritocracy trap needs a measure to enforce itself. Metric fixation supplies one and lends it the authority of arithmetic.
Related Patterns
- Pattern 39 — Average User Fallacy
- Pattern 11 — Burnout Threshold
- Pattern 49 — Environmental Weathering
- Pattern 02 — Spiky Profiles
Related Principles
- When You Measure, Include the Measurer
- Broken Systems, Not Broken People
- Care Work Makes All Other Work Possible
- Foreground Complexity as the Baseline