Stop Running With Sharp Metrics
summary
Argues that metrics must map to outcomes and warns specific harms like a public SLA dashboard that names engineers creating perverse incentives and rushed fixes. Recommends concrete practices: GQM (goals questions metrics) mapping, the SEET test (matters to customers or drives decisions), watch deployment‑frequency as a vanity metric, and balance cycle time, incident counts, customer satisfaction and SLAs in quarterly QBR visuals so tradeoffs are visible and metrics get retired when they stop helping.
AI-written summary · about
chapters (27)
- 0:00Intro
- 0:48About Julia Wester
- 2:11Types of business analytics
- 3:43Metrics are important
- 4:22Risk
- 5:12Outline
- 5:45Antipatterns
- 7:03Order of Operations
- 7:23Measure Everything
- 8:09Vanity Metrics
- 9:23Deployment Frequency
- 10:16Shame
- 12:40Breach Lists
- 13:42Lack of Awareness of the Bigger Picture
- 15:31Reading Metrics Like Targets
- 17:24Knowing How To Do It
- 17:41Know Your Goal
- 17:53GQM Goals Questions
- 19:05So Wet Test
- 19:31Deploy Frequency
- 22:51Involve Your Team
- 23:36Use Deploy Frequency
- 24:37Balance Quadrant
- 26:46Example
- 29:45Understanding your situation
- 30:53Recap
- 31:56Slides
related talks
01h 01mFinding Metrics that matter and using them safely
30mMetrics for good, not evil
13mObservability that matters
14mWhat dashboards don't tell you
23mYou Get What You Measure—Why Metrics Are Important
18mHow not to practice observability
42mThe Fault in Our Metrics: Rethinking How We Measure Detection & Response
05mMetrics That Matter