T20

Goodhart's Law ('when a measure becomes a target, it ceases to be a good measure') is widely cited in economics and management science but less applied to GTM measurement design. The contestation is practical: GTM leaders cannot measure everything, so they must choose metrics, and any chosen metric will be gamed to some degree. The prescriptive implication — measure leading indicators and rotate metrics to prevent gaming — is organizationally disruptive. Leaders who have built incentive structures around existing metrics resist changing them, even when they observe Goodharting in practice.

The GTM system is reflexive (T19): the act of measurement changes the behavior being measured. Specific to CRM: (a) if stage conversion rates are measured and compensated, reps will optimize for stage advancement over customer fit, producing a pipeline filled with poorly-qualified opportunities; (b) if 'activities' (calls made, emails sent) are measured, reps will prioritize high-volume low-quality activity over selective high-quality engagement; (c) if last-touch attribution is used (T6), campaigns that touch the bottom of the funnel will be over-invested, and brand programs that touch the top will be starved. The Goodhart dynamic: every metric that is both measurable and compensated will be optimized toward the metric rather than toward the underlying value it was meant to proxy. The substrate (CRM, attribution model, incentive design) is not a neutral observer — it actively shapes the behavior of the GTM system it measures.

• Classic B2B case: call metrics at a large insurance company led to reps making 100+ short calls to game the metric; removing the metric and replacing it with revenue-per-account led to longer, higher-quality conversations and better retention
• HubSpot internal research: companies that measured 'MQLs' as a primary marketing metric systematically over-weighted activities that produced high MQL volume; switching to 'pipeline-qualified' as the primary metric changed investment allocation significantly
• Salesforce study of pipeline accuracy: companies that incentivize opportunity advancement (stage-based commission) had 35% lower forecast accuracy than companies that incentivize close revenue — the incentive structure corrupted the predictive signal
• Bain survey of B2B CMOs: 60% reported that their primary metrics were lagging indicators that incentivized the wrong behaviors; most knew the metrics were wrong but felt unable to change them

• Rotating metrics to prevent Goodharting creates organizational instability — reps cannot optimize if the goalposts move quarterly
• Some metrics are resistant to gaming in practice because gaming requires effort that exceeds the reward (e.g., closed ARR is hard to Goodhart because it requires actual customer payment)
• Data observability tools (BI dashboards, revenue intelligence platforms) increasingly catch anomalies in CRM data that would previously have enabled Goodharting undetected

Audit your current GTM metric stack for Goodhart risk: for each metric, ask 'what behavior does optimizing this metric incentivize?' and compare to 'what behavior do we actually want?' Where they diverge, you have a Goodhart problem. The fix is not to remove measurement but to use composite metrics that are harder to game in isolation, lag metrics that capture ultimate economic outcomes, and qualitative calibration (deal review, win/loss analysis) alongside quantitative dashboards. As an operator: never compensate on a single leading metric without a lagging backstop that captures the economic outcome you actually care about.