Skip to content

Team Performance Science Guide · Books · 8 of 22

Rethinking Performance Measurement

Marshall W. Meyer · 2002

Org/strategy measurement (critique)·Stance: Ally

Ally — the most psychometrically sophisticated critic in the cluster

File: ~/Desktop/Performance Books/Rethinking Performance Measurement - Beyond the Balanced Scorecard ...pdf Tradition: Org/strategy measurement (critique) Stance toward Performix: Ally — the most psychometrically sophisticated critic in the cluster Read depth: Intro + Ch.1–2 close-read via keyword extraction

Core claim (1 line)

Performance measurement fails because what we want to measure (long-term viability) and what we can measure (current proxies) are not the same — and the Balanced Scorecard doesn't fix this, because it gives "no guidance on how to combine dissimilar measures into an overall appraisal of performance."

What the book actually says (grounded)

  • Want-vs-can gap: "the performance we want to measure (long-term cash flows, long-term viability) and the performance we can measure (current cash flows, customer satisfaction, etc.) are not the same."
  • The "running down" of measures (the standout insight): "the tendency of almost all measures to lose variance and hence the capacity to discriminate between good and bad performance." Measures decay as discriminators; "running down is attenuated in turbulent environments."
  • Too many measures: "firms are swamped with measures … fifty to sixty top-level measures"; the IMA finds a large share of accountants rate their measures "poor" — and "do not experience these changes as improvements."
  • Seven purposes (often in conflict): look ahead · look back · motivate · compensate · roll up · cascade down · compare — the roll-up/cascade/compare (org artifacts) pull against motivate/compensate.
  • Proposed fix — ABPA (activity-based profitability analysis): "estimates the revenue consequences of each activity performed for the customer," letting firms "discriminate between activities that are ultimately profitable and those that are not."

Contrast with the Performix thesis

Performix commitmentMeyer's positionVerdict
Diagnostic, not laundry-listAttacks measure-sprawl head-on ("swamped with measures")Strong ally
CAMS binding constraintNo binding-constraint model, but the want/can gap motivates measuring the right latent thingAlly (philosophical)
Psychometric-first / AI-as-consumer"Running down" = loss of variance/discrimination — this is a psychometric property (item information / floor-ceiling) stated in business languageDirect ally — validates our IRT/uncertainty stance and the need to monitor a measure's discriminating power
Adaptive narrowing + VOI"Too many measures"; measure what discriminatesAlly
Three leversABPA = translate activity → value ≈ our "ops → performance value" design lawAlly ([[project_performix_translates_ops_to_performance_value]])

Extractables

  • Construct/measure: measure "running down" / discrimination decay → a substrate health-check: monitor whether a CAMS item still discriminates (variance not collapsing). Strong fit with index doctrine + reincarnation item lifecycle.
  • Concept: the want-vs-can gap → frame for why we measure latent CAMS constructs (the want) via observable indicators (the can), with stated uncertainty.
  • Tool / PFX card idea: ABPA-style activity→value translation as a pattern for the value-stack (don't surface raw ops).
  • Substrate ingest candidate: yes — the seven-purposes taxonomy + running-down concept as priors for measurement-design/health.

Content seeds

  • Article angle: "Your metrics are 'running down' — why good measures quietly stop telling you anything." (variance decay; why dashboards go stale).
  • Guide entry: Part VII (measurement) — the want/can gap + measure-health; cross-ref Part I (refuse the laundry list).

Library record

  • Canonical id: (mint per library SPEC; tag performix) · Path A/B: not yet ingested
  • Already in PA library? Not found. Aligned → move to PA inbox (batch with SDT/Ericsson/Marr).

Grounds these Guide parts

Library record

No canonical library id minted yet. The PeopleAnalyst library registry stays the identity layer; this profile will carry the id once the work is ingested.

← All books