Observability has become a board-level issue as cyber incidents trigger disclosure deadlines, but the standard data foundation of metrics, events, logs, and traces (MELT) is no longer sufficient. A sponsored report from The Register, based on NETSCOUT research, argues that organizations are collecting more data than ever yet still lack conclusive insight. Executives are left trusting their gut despite rising spending on telemetry.
The numbers are stark. While 96 percent of organizations use metrics and logs, 82 percent report visibility gaps and 96 percent say they lack sufficient data to determine root cause during incidents. Visibility often disappears at boundaries: 58 percent lose sight between on-premises and cloud, 51 percent at the edge, and 39 percent in service-to-service interactions. Downtime costs are high, with 42 percent estimating $500,000 to $999,000 per hour.
The stakes grow higher with AI-driven operations. Autonomous agents need continuous, unsampled, forensic-grade data to make reliable decisions, but telemetry is often sampled or aggregated. Only 41 percent of organizations describe AI-assisted insights as very or extremely consistent, and 38 percent admit they lack forensic-grade data to validate automated actions. As NETSCOUT's Jack Callahan puts it, feeding an agent partial or sampled data risks "scaling that uncertainty really quickly."
The report's proposed fix is to enrich MELT data with network context rather than replace existing observability platforms. That approach, however, comes from a vendor-sponsored feature, so readers should weigh the commercial interest behind the recommendation.