By Anthony Woods, Co-Founder and former CTO of Grafana Labs. Experienced engineering leader supporting high-velocity software delivery.
getty
When most people hear “observability,” they think of on-call rotations, alerts and dashboards for SREs. That narrow view is changing. Over the past few years, observability tools and the practices around them have moved beyond operations teams and into the rest of the organization: finance, sales, product and even the boardroom.
At its core, observability is the practice of understanding how systems work. It’s about turning black boxes into glass boxes so you can see what’s happening inside. When something unexpected occurs, that visibility allows you to trace signals and pinpoint the root cause.
Traditionally, we’ve applied that thinking to software systems. But the same principle applies to the way organizations operate. It’s not about turning everyone into a data scientist; it’s about visibility. If you can measure something, you can improve it.
Two trends make this possible. First, teams are becoming more data-driven across the company as organizations move beyond departmental solutions toward enterprise-wide data democratization, with a data-driven culture emerging as one of the top organizational priorities of 2026. Second, AI is lowering the barrier to using that data. But neither replaces human judgment or technical expertise; instead, they change where and how it’s applied.
Most companies collect far more data than they use. Historically, that data sat in silos: a finance spreadsheet here, a CRM report there, application logs elsewhere. What’s changing now is that organizations are bringing those sources together. Centralized analytics and BI teams are often the first to stitch disparate data into a single view. When that happens, finance leaders gain operational context; product teams correlate customer behavior with incidents; sales leaders track seller productivity over time.
We’re already seeing real examples. Some organizations have replaced traditional BI tools with flexible platforms that create shared views across teams. Internally at Grafana Labs, we’ve seen engineering telemetry combined with business process data to understand developer productivity and release cadence, not just latency or error rates. Those are business metrics explored through observability tools. When you can see a process, you can measure it and reduce friction.
AI accelerates this shift by making data accessible through natural language. The goal is simple: someone should be able to ask, “Show me seller performance over the last quarter,” and get a useful dashboard.
But AI is only as good as the data and metadata behind it. Raw data looks like noise to a language model. The missing ingredient is context.
That’s where knowledge graphs and enriched metadata come in. If you take disparate records and add the data-about-the-data (what fields mean, how events map to business processes and who owns a dataset), you create a semantic layer that both humans and machines can understand. That structure turns AI from a novelty into a practical interface for data.
This evolution doesn’t eliminate the need for technical talent; it shifts it. Rather than centralizing expertise in a single platform org, knowledge and structure move closer to where data is produced. Analysts and engineers become architects of shared meaning; the rest of the company becomes the consumer.
When people see how meaningful data can be, they start building solutions for themselves. At Grafana, we’re seeing non-engineers build dashboards that answer role-specific questions, and those small wins often spread across teams. Adoption grows through usefulness, not mandates.
This is where observability becomes powerful beyond engineering. In a software system, observability helps you understand why something broke. In a business system, it can help you understand why something worked.
Within Sales, for example, observability can reveal not just pipeline health but patterns behind high-performing teams or individuals. When you can see the signals behind success, activity patterns, timing and collaboration across roles, those behaviors become repeatable.
Observability doesn’t just diagnose failure. It helps organizations learn from success.
Of course, there are real risks. AI hallucinations are not hypothetical; models sometimes produce confident but incorrect outputs, and decisions based on them can be costly. There’s also the classic “trash in, trash out” problem: collecting everything indiscriminately increases costs without guaranteeing insight. Democratization without guardrails can also lead to conflicting metrics and inconsistent definitions.
The right approach balances empowerment with discipline. Provide tools that let people ask questions, but invest in the plumbing that makes answers trustworthy: discoverable data, clear metadata and shared definitions. Treat observability as an organizational capability, one that connects engineering rigor with business context.
If you measure what matters and structure your data so machines can understand it, observability becomes more than a tool for fixing outages. The companies that win won’t be the ones with the most data, but the ones whose data can be understood by both people and machines.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。