[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"glossary-proxy-metric::en":3,"gloss-cluster-proxy-metric::en":23,"gloss-next-proxy-metric::en":9},{"slug":4,"category":5,"name":6,"definition":7,"meta_desc":8,"faq":9,"schema_markup":9,"related":10},"proxy-metric","analytics","Proxy Metric","A proxy metric is something you can count standing in for something you care about but cannot count. Nobody can instrument the customer got value from us this week, so a product counts weekly active accounts, or documents created, or reports run. A support tool counts resolved sessions because it cannot see whether the person's problem actually went away. The substitution is unavoidable and mostly reasonable; the mistake is forgetting that it happened and arguing about the proxy as though it were the thing. Every proxy carries a gap between what it counts and what it represents, and most disputes about a dashboard are really disputes about that gap. Treating a proxy as a hypothesis makes it tractable, because hypotheses can be checked. The claim is that movement in the countable thing implies movement in the thing that matters, and you can test it once, expensively, against something slower and more direct: interviews, renewal outcomes for accounts that scored high and low on the proxy, or a survey asked of both groups. The result is not permanent. A proxy is calibrated against a particular version of a product, and a redesign can sever the link without a line of tracking code changing — a metric counting reports run stops meaning engagement the week reports start being scheduled automatically. Two practices keep proxies honest. Report the proxy under the name of the proxy rather than the name of the outcome, so that documents created never quietly appears in a deck as customer value. And when a proxy is used as a target, expect it to decay faster, because a measure under pressure loses its relationship to the underlying thing sooner than one merely observed. It is also worth writing down what the proxy is known to miss at the moment it is adopted, because that list is obvious on the day of the decision and invisible a year later, when the only people who remember the caveat have moved teams and the number has acquired the authority of long use.","A proxy metric counts what you can instead of what you mean. How to test the link, why redesigns break it, and why targets make proxies decay faster.",null,[11,14,17,20],{"slug":12,"name":13},"active-user","Active User (DAU, WAU, MAU)",{"slug":15,"name":16},"guardrail-metric","Guardrail Metric",{"slug":18,"name":19},"north-star-metric","North Star Metric",{"slug":21,"name":22},"retention-curve","Retention Curve",[24,28,29,32,35,38,42,45,48,51,54,55],{"slug":25,"category":5,"name":26,"updated_at":27},"ab-testing","A\u002FB Testing","2026-09-08T21:32:36+00:00",{"slug":12,"category":5,"name":13,"updated_at":27},{"slug":30,"category":5,"name":31,"updated_at":27},"autocapture","Autocapture",{"slug":33,"category":5,"name":34,"updated_at":27},"cost-per-resolution","Cost per Resolution",{"slug":36,"category":5,"name":37,"updated_at":27},"customer-data-platform","Customer Data Platform (CDP)",{"slug":39,"category":5,"name":40,"updated_at":41},"data-sampling","Data Sampling","2026-09-08T21:32:37+00:00",{"slug":43,"category":5,"name":44,"updated_at":27},"deflection-rate","Deflection Rate",{"slug":46,"category":5,"name":47,"updated_at":41},"double-counting","Double Counting",{"slug":49,"category":5,"name":50,"updated_at":41},"feedback-signal","Feedback Signal",{"slug":52,"category":5,"name":53,"updated_at":41},"goodharts-law","Goodhart's Law",{"slug":15,"category":5,"name":16,"updated_at":27},{"slug":56,"category":5,"name":57,"updated_at":27},"identity-resolution","Identity Resolution"]