Proxy Metrics in Experiments

Proxy Metrics in Experiments

Proxy metrics in experiments are faster, more frequent, and less noisy signals used to make experiment decisions when they are logically and historically linked to a slower downstream KPI.

Key points

  • Statsig argues that revenue can be too sparse and noisy to use as the primary decision metric for every experiment, even when it remains the ultimate business KPI [src-031].
  • A useful proxy sits up-funnel from the target outcome, fires more often, and has a stable historical correlation with the downstream KPI [src-031].
  • Proxy metrics should be less exposed to external shocks such as holidays or marketing pulses and should reduce noise enough to create faster evidence [src-031].
  • The article’s example is using clicks on “Add to cart” as a faster signal for purchase conversion when historical analysis supports the relationship [src-031].
  • The downstream KPI should not disappear; Statsig recommends keeping it as a guardrail while deciding the experiment on the validated proxy [src-031].

Related entities

Related concepts

Source references

  • [src-031] Yuzheng Sun — “Speeding up A/B tests with discipline” (2025-06-24)

Robin Cartier perspective

This page is part of Robin Cartier's working AI knowledge graph: a practical research layer for production AI, recommendation systems, experimentation, GEO, and agentic web readiness.

The useful next step is to connect this concept back to applied product leadership and operating models.

Recommended next

Keep reading from this thread

From 491 indexed pages and articles.

  1. Wiki concept Experiment Statistical Power The ability of a test to detect a real effect when that effect exists. Key points Statsig's parallel-testing article Related by faster
  2. Wiki concept Experiment Variance Reduction The use of statistical and assignment methods that lower noise in experiment readouts, narrowing confidence intervals so fewer samples Related by metrics
  3. Insight Recommendation Systems in Production How recommendation systems become production decisioning systems through signals, ranking, constraints, feedback loops, and experimentation Related by metrics