Enterprises spend USD $2.5 million on logging tools

Dynatrace has published research showing that enterprises spend nearly USD $2.5 million a year on logging tools while excluding most log data from analysis. The study surveyed 450 senior technology leaders at large companies.

The findings point to growing strain on traditional log management as AI workloads increase the volume of operational data companies generate and need to assess. Log and telemetry volumes have risen by 93% over the past 12 months, according to the research.

Organisations now use an average of seven tools to manage logs and telemetry. That fragmentation leaves teams comparing and correlating information manually across separate systems, which respondents linked to slower analysis and operational delays.

Four in five respondents said the challenge of turning telemetry into actionable insights is hurting customer experience and delaying AI initiatives. Nearly half said they discard or do not collect logs, with organisations excluding an average of 86% of log data from ingestion, storage or analysis to control costs and work around system limits.

Rising costs

Logging costs cover ingestion, management, storage, indexing, rehydration and querying. Even with that outlay, many teams limit the volume of data they keep, creating a gap between the visibility AI systems require and the economics of existing logging approaches.

This tension matters because logs remain a central source of information for monitoring system behaviour, validating outputs and investigating faults or security issues in AI-driven environments. As more AI applications move into production, the telemetry they generate is putting increasing pressure on tools designed for earlier, less complex workloads.

Dynatrace also found that roughly a third of organisations are paying for redundant or underused observability features. More than a quarter said engineering time is spent maintaining multiple tools across environments rather than supporting AI systems in live use.

Platform shift

Nearly three-quarters of respondents said AI workloads require a platform-based approach to log management. A further 81% said log ingestion and processing must be open and automated to support real-time analysis.

The research suggests companies want to combine logs with other forms of telemetry, including metrics, traces and events, rather than manage each data type in isolation. That reflects a broader shift in IT operations as businesses try to reduce the number of separate monitoring tools they use.

Large enterprises in the survey had annual revenues of USD $750 million or more, suggesting the issue is not confined to smaller organisations with limited technology budgets. Instead, the results indicate that scale itself is creating operational and financial pressure as AI systems produce more data than existing logging models can handle economically.

Mala Pillutla, Vice President of Log Management at Dynatrace, said the current setup is proving inadequate for AI-heavy environments. "AI is accelerating enterprise innovation, but most logging systems were never built for the scale, speed, or complexity of AI-driven environments," Pillutla said.

"As AI agents operate probabilistically, treating logs, metrics, traces, and events as separate signals is no longer viable. To make AI systems reliable and trustworthy, organisations need a unified, intelligent approach that brings all telemetry together in real time, enriched with deep context to drive confident decisions," Pillutla said.

The figures add to a growing debate in the technology industry over the cost of observability, particularly as AI increases demand for system monitoring, auditability and performance tracking. For many enterprises, the challenge is no longer simply collecting data, but deciding what to keep, what to discard and how to make sense of it without adding further complexity.

One of the clearest findings is that companies are already making trade-offs. Despite spending nearly USD $2.5 million a year on logging, respondents said they still exclude most of their log data, with an average of 86% left out to manage costs and system limitations.