Uploaded August 2026 | Updated September 2026, 2 weeks ago
PointsBet runs a fast-paced, high-stakes wagering platform across Australia and Ontario, Canada — where every second of incident detection delay has real revenue and customer impact. Staff SRE Adam Dagdelen walks through how PointsBet's infrastructure team moved off a fragmented, vendor-locked observability stack (Application Insights, Datadog, and Azure Managed Grafana dashboards, bolted together with PagerDuty and Jira) and rebuilt around Observability 2.0: one platform, one protocol, full cost attribution, all based on Grafana Cloud
In this talk, Adam covers:
• Why scattered tooling and vendor lock-in became the org's biggest blind spot
• Using OpenTelemetry as the "exit ramp" to decouple instrumentation from any single vendor
• A dual-export migration strategy that avoided a hard cutover ("no flag day")
• Scaling the OpenTelemetry Collector and Alloy receiver with a GitOps-based ownership model
• Golden-path alert and dashboard templates that sped up team onboarding
• Building an observability champions network to drive org-wide adoption
• An estimated 35% reduction in observability spend
• Where AI — plain-language PromQL and first-pass root cause investigations — fits into the next phase
00:00 Introduction — who is PointsBet
01:46 Why observability matters for a wagering platform
03:13 The cracks: scattered tooling, vendor lock-in, obscure costing
04:40 Defining Observability 2.0
05:17 Why Grafana Cloud
06:29 OpenTelemetry as the exit ramp
09:21 Labeling strategy and attribution
10:20 The dual-export migration approach
10:42 Scaling the OpenTelemetry Collector
13:06 Results: cost savings and cost attribution
14:24 The hardest part: culture and adoption
15:16 Observability champions
16:44 Golden-path alerts and dashboards
18:21 AI as the newest adoption lever
20:50 What's next for PointsBet
Explore Grafana Cloud → grafana.com/products/cloud
PointsBet runs a fast-paced, high-stakes wagering platform across Australia and Ontario, Canada — where every second of incident detection delay has real revenue and customer impact. Staff SRE Adam Dagdelen walks through how PointsBet's infrastructure team moved off a fragmented, vendor-locked observability stack (Application Insights, Datadog, and Azure Managed Grafana dashboards, bolted together with PagerDuty and Jira) and rebuilt around Observability 2.0: one platform, one protocol, full cost attribution, all based on Grafana Cloud
In this talk, Adam covers:
• Why scattered tooling and vendor lock-in became the org's biggest blind spot
• Using OpenTelemetry as the "exit ramp" to decouple instrumentation from any single vendor
• A dual-export migration strategy that avoided a hard cutover ("no flag day")
• Scaling the OpenTelemetry Collector and Alloy receiver with a GitOps-based ownership model
• Golden-path alert and dashboard templates that sped up team onboarding
• Building an observability champions network to drive org-wide adoption
• An estimated 35% reduction in observability spend
• Where AI — plain-language PromQL and first-pass root cause investigations — fits into the next phase
00:00 Introduction — who is PointsBet
01:46 Why observability matters for a wagering platform
03:13 The cracks: scattered tooling, vendor lock-in, obscure costing
04:40 Defining Observability 2.0
05:17 Why Grafana Cloud
06:29 OpenTelemetry as the exit ramp
09:21 Labeling strategy and attribution
10:20 The dual-export migration approach
10:42 Scaling the OpenTelemetry Collector
13:06 Results: cost savings and cost attribution
14:24 The hardest part: culture and adoption
15:16 Observability champions
16:44 Golden-path alerts and dashboards
18:21 AI as the newest adoption lever
20:50 What's next for PointsBet
Explore Grafana Cloud → grafana.com/products/cloud
