Enterprises are spending big on observability. But most of the data they’re paying for is never used. And worse, it’s making troubleshooting harder.
At Sawmills, we just released the 2025 State of Observability and Telemetry Report, a survey of senior DevOps and engineering leaders across the US and EU. The results paint a clear picture: observability is overwhelmed with redundant and irrelevant telemetry data that’s blowing up costs and slowing down debugging.
Key Findings at a Glance
- Only 13% of collected telemetry is used. 84% of companies use less than a quarter of the data they ingest. The rest is expensive noise.
- Costs are exploding. The average enterprise now spends $905,000 annually on observability, with many topping $1M.
- Tool sprawl is prevelant. 86% of companies run two or more observability platforms, with Datadog, New Relic, and Grafana/Loki/Prometheus leading the pack.
- AI adoption is on the rise. 63% of companies are interested in AI agents to optimize telemetry and reduce costs; 60% want AI copilots for SRE teams to accelerate incident response.
What This Means for Engineering Teams
The old mantra of “collect everything” is problematic. Every extra log line, trace span, or high-cardinality metric isn’t just cluttering dashboards — it’s driving up costs and hiding the signals that matter. When engineers spend more time sifting through noise than solving problems, mean time to resolution (MTTR) suffers, and so does reliability.
The Shift to Smart Telemetry
Observability is no longer just about visibility. It’s about control. Teams that continue down the “collect everything” path will keep paying more while seeing less. Teams that embrace smarter telemetry pipelines will restore clarity, speed up resolution, and finally align observability costs with business value.
That’s where Sawmills comes in. By analyzing telemetry in real-time, a smart telemetry platform can filter out noise, highlight anomalies, and improve the signal-to-noise ratio, helping businesses cut costs while giving engineers more clarity. It’s no surprise that nearly two-thirds of the companies we surveyed are exploring AI for telemetry management.
Observability isn’t just a tooling challenge. It’s a data challenge. The companies that solve their data waste problem will not only spend less — they’ll resolve incidents faster and operate with more reliability.
