Every modern organisation is producing an extraordinary volume of telemetry, with applications, infrastructure, cloud services and user interactions constantly generating signals about how digital systems are performing. Yet, having access to more telemetry and data has not necessarily given teams greater clarity. Instead, it has created a new challenge of identifying which signals actually matter and making sure they reach the right teams and systems at the right time.
As technology environments become more dynamic and distributed, that challenge is becoming harder to ignore. Teams can find themselves spending vast amounts of time searching across multiple disconnected sources to understand an issue or outage while the disruption continues. When relevant signals are buried among large quantities of low-value data, determining what course of action should be taken to solve an issue can take much longer, not to mention the costs associated with storing such huge volumes of telemetry.
That means the focus needs to shift from how much telemetry and data organisations can collect to how effectively they manage it from the outset. Building intelligence into the telemetry lifecycle earlier can reduce noise before it travels downstream and ensure that what remains is relevant and actionable. For teams under pressure to move faster while also managing increasingly complex environments, getting this right at the source becomes critical.
Better outcomes start with better signals
It’s clear that the challenge isn’t that organisations are short of telemetry and data; it’s that they’re short of good quality signals. As AI-driven capabilities become deeply embedded in software delivery and operations, the quality of observability telemetry becomes even more important. When the telemetry is collected and unnecessary telemetry is filtered out, employees end up spending more time managing the telemetry than learning from it.
Part of this issue is fragmentation, when telemetry is stored across several tools and agents, each of which generates signals with somewhat varied forms. During an outage, teams are forced to stitch together inconsistent signals across tools, losing valuable time while the business impact grows.
The financial part comes next. Large-scale telemetry and data processing and storage are expensive, and the costs are only rising. Organisations are spending a significant amount of their resources on processing and storing data that only adds noise, not insights. In the traditional sense, this isn’t a technology problem. It’s a telemetry and data quality concern, and the longer it’s ignored, the worse it usually gets.
Shaping telemetry before it travels downstream
Moving intelligence upstream means filtering, enriching and routing meaningful telemetry before it reaches downstream systems. Increasingly, organisations are starting to think about how to address these fragmentation and financial challenges. Teams are beginning to collect, process, and route relevant observability telemetry in real-time before it is ingested into downstream systems. As a result, there is a less volume to handle, reduced expenses and cleaner telemetry to deal with since only the relevant signals are sent downstream.
This is becoming increasingly feasible at scale thanks to open standard such as OpenTelemtry. For example, teams can gather telemetry consistently across multiple environments without being dependant on a particular vendor or set of tools. When working with intricate, distributed systems that span several clouds, security and storage platforms, that type of flexibility is crucial. Furthermore, as environments continue to evolve, it makes adapting to them that much easier.
Giving AI the context it needs
AI systems are only as robust as the telemetry and data that powers them. They depend on accurate, contextualised telemetry to generate insights and automate decisions. Teams are not only slowed down by incomplete or cluttered telemetry, but it’s this simple: poor quality telemetry leads to poor quality insights. When those insights are applied to large-scale automated decision-making, the margin for error drastically lowers.
Another aspect that is frequently disregarded is trust. The teams in charge of these automated systems must be able to explain what happened and why these decisions are being made by these systems about performance or security. If the information supporting those choices is inconsistent, it becomes extremely challenging. This matters far beyond just the technical teams. Senior management and board members are being increasingly asked to support AI-driven choices, and this becomes a much simpler discussion when the telemetry and data behind the insights is credible and trustworthy.
The operational advantages are also evident. Teams can work quicker if they aren’t continuously battling erratic telemetry and data. They spend less time on manual research, detect issues early, and automate with greater confidence. A strong telemetry and data foundation and a unified, context-driven platform not only enhance daily operations but also increases the successes of the AI-powered capabilities the organisation invests in over time.
Rethinking what good observability looks like
Perhaps the bigger shift here is one of mindset. For years, the default assumption has been that retaining more telemetry creates more visibility into a business. However, as digital environments grow in scale and complexity, and AI is thrown into the mix, that assumption becomes harder to sustain, and the ability to collect everything is no longer the same as the ability to understand everything.
This changes what good observability looks like. Success will increasingly depend on how deliberately organisations design the journey from signal to decision. That puts greater emphasis on making choices at the point where telemetry is created, rather than relying on downstream systems to make sense of everything later.
It also changes the role observability plays within the wider technology strategy. As software becomes more dynamic and automated, observability cannot simply be a record of what has already happened. Instead, it needs to provide the foundation for understanding what is happening now, what matters to the business and what should ultimately happen next.
The winning organisations will be those turning observability into a source of intelligence itself – one that gives both people and AI the clarity to act when it matters.
Mala Pillutla
Mala Pillutla is Vice President – Observability & Security at Dynatrace where she leads the strategy and execution of specialist teams worldwide. Drawing on extensive experience in enterprise software, observability and security, Mala focuses on helping organisations navigate complex cloud and IT environments with scalable solutions shaped around customer needs.
