---
tags: ["synthesis", "measurement", "kpis", "ai-adoption"]
articles: ["a127", "a130", "a121"]
synthesis: true
id: "cross-metrics-mislead-managers"
sources: ["tail2"]
type: "synthesis"
sourceVaultSlug: "hbr-seg-tail2"
originDay: 2
articleStem: "hbr-seg-tail2"
sourceUrl: "(unified vault: 14 sources)"
sourceTitle: "HBR — Tail Ⅱ · Founders, PE, 2025 items, industry/security/ops (#118–131)"
---
## The measurement trap, three ways

Three articles independently warn that the numbers on a leader's dashboard can point in exactly the wrong direction.

- **A127 — usage is a false proxy:** [[claim-usage-not-buy-in]] and [[concept-performative-ai-usage]] show that high AI utilization can be *fear-driven compliance*. The most jarring finding — [[contrarian-anxiety-drives-usage]] — is that rising usage plus high angst is a warning sign, not a win.
- **A130 — local wins mask global failure:** [[contrarian-local-success-global-failure]] documents a retailer whose siloed AI wins (15% fewer stockouts, 40% faster response, 25% higher email opens) coincided with flat satisfaction and lost market share. Optimizing local metrics reverses corporate performance ([[claim-ai-reinforces-silos]]).
- **A121 — lagging vs. leading:** top CEOs track [[concept-leading-indicators-of-focus]] (pipeline, wins, capacity) rather than only revenue/EBITDA, and convert status meetings into decision forums.

## The shared prescription

All three say: pair the easy metric with a truer signal. A127 pairs telemetry with psychological-safety and angst measures ([[action-pair-metrics-with-safety-signals]]); A130 replaces function metrics with [[concept-shared-cross-functional-kpis]]; A121 replaces lagging with leading indicators. The unifying principle is that activity is not impact, and that a metric optimized in isolation ([[prereq-adoption-telemetry]]) becomes a lie. This is the measurement corollary of the psychology theme in [[cross-leadership-as-psychology]].