[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"glossary-guardrail-metric::en":3,"gloss-cluster-guardrail-metric::en":26,"gloss-next-guardrail-metric::en":9},{"slug":4,"category":5,"name":6,"definition":7,"meta_desc":8,"faq":9,"schema_markup":9,"related":10},"guardrail-metric","analytics","Guardrail Metric","A guardrail metric is a measure you watch not to improve but to make sure it does not get worse. Every optimisation target can be moved by means nobody intended: signups rise when the pricing page hides the price, activation improves when a step is skipped that prevented refunds later, engagement climbs when notifications become intrusive. Guardrails are the counterweights declared in advance, so a change that wins on the primary metric and loses somewhere important is recorded as a loss rather than a success. Useful guardrails are chosen per experiment from the plausible side effects of the specific change, plus a small standing set the whole team watches: refund or chargeback rate, support contact rate, unsubscribe rate, page latency and error rate. They are usually evaluated differently from the primary metric — you are not trying to prove an improvement, only to detect a degradation large enough to matter, so a threshold agreed beforehand is more useful than a significance test that a low-traffic guardrail will rarely clear. Two habits keep them honest. Write them down before the experiment starts, because a guardrail chosen afterwards tends to be one that happens to look fine. And define the action in advance: which guardrail movements block a rollout, which require a follow-up, and who decides. Without that, a guardrail is a chart someone screenshots after the decision has already been made.","A guardrail metric is watched to catch damage rather than to improve — how to pick them per experiment and why the response must be agreed before launch.",null,[11,14,17,20,23],{"slug":12,"name":13},"ab-testing","A\u002FB Testing",{"slug":15,"name":16},"error-budget","Error Budget",{"slug":18,"name":19},"north-star-metric","North Star Metric",{"slug":21,"name":22},"product-analytics","Product Analytics",{"slug":24,"name":25},"statistical-significance","Statistical Significance",[27,29,32,36,39,42,45,48,51,54,57,60],{"slug":12,"category":5,"name":13,"updated_at":28},"2026-08-24T02:46:38+00:00",{"slug":30,"category":5,"name":31,"updated_at":28},"active-user","Active User (DAU, WAU, MAU)",{"slug":33,"category":5,"name":34,"updated_at":35},"autocapture","Autocapture","2026-08-24T02:46:37+00:00",{"slug":37,"category":5,"name":38,"updated_at":28},"cost-per-resolution","Cost per Resolution",{"slug":40,"category":5,"name":41,"updated_at":35},"customer-data-platform","Customer Data Platform (CDP)",{"slug":43,"category":5,"name":44,"updated_at":28},"deflection-rate","Deflection Rate",{"slug":46,"category":5,"name":47,"updated_at":35},"identity-resolution","Identity Resolution",{"slug":49,"category":5,"name":50,"updated_at":35},"multi-touch-attribution","Multi-Touch Attribution",{"slug":52,"category":5,"name":53,"updated_at":28},"novelty-effect","Novelty Effect",{"slug":55,"category":5,"name":56,"updated_at":35},"retention-curve","Retention Curve",{"slug":58,"category":5,"name":59,"updated_at":28},"sample-ratio-mismatch","Sample Ratio Mismatch (SRM)",{"slug":61,"category":5,"name":62,"updated_at":28},"seat-utilization","Seat Utilization"]