How SREs Found More than $100 Million Using Failed Customer Interactions
summary
PayPal shifts from inward payment metrics to tracking Failed Customer Interactions (FCIs) by tagging requests with a correlation ID and surfacing HTTP 500s in dashboards that tie failures to owning services. Real-time alerts and automated rollbacks let teams fix issues on the spot; systemic FCIs fall ~95%, daily availability climbs to four nines, calls drop ~$20M, and revenue rises about 1%—roughly a hundred million dollars—while region and flow-specific drills (Turkmenistan signup example) guide targeted fixes.
AI-written summary · about
chapters (20)
- 0:00Introduction
- 0:30PayPal
- 2:12Payment Volume
- 2:38Site Reliability Engineer
- 3:13How we measure Availability
- 5:17Whats wrong with that model
- 7:08New measurement needed
- 8:17What is an MCI
- 9:23What is a correlation ID
- 11:00Why correlation ID is important
- 12:10Heatmap
- 12:59Service Ownership
- 14:03Regions
- 14:33Turkmenistan
- 16:47Realtime FCI Demo
- 17:17What makes this whole thing successful
- 19:55Death by a thousand cuts
- 20:46Awesomeness chart
- 21:57Results
- 24:13Questions
related talks
38mLeveraging Data to Unlock Business Value
34mData for dollars - How fan insights drive revenue
07mDynamic Decision Making at HMRC - Moving to Data as a Product During a Pandemic
13mFinTech pt.3
15mUsing Statistical Techniques to Automatically Detect Game-Breaking Issues
28mHow SRE Changes the perspective of Data Centric Organization and Data....
01h 04mLeveraging Open Data to Gain Advantage on the Competition
41mData DevOps is Real!