CX Observability products allow organizations to monitor their end-to-end customer experience across networks, interactions and agents, and quickly act to maintain service.
Originally published on No Jitter, April 25, 2025 and authored by Nicolas De Kouchkovsky
The Rise of CX Observability in Service Monitoring
CX Observability products allow organizations to monitor their end-to-end customer experience across networks, interactions and agents, and quickly act to maintain service.
In today's $300 billion contact center market, the difference between market leaders and laggards increasingly comes down to one factor: their ability to extract actionable intelligence from complex cloud operations. As contact centers migrate to the cloud and embrace AI, the complexity of their technology stacks has skyrocketed. Managing such distributed environments and agent workforces requires more than traditional monitoring tools—it demands a deeper understanding of how every component of the customer experience ecosystem interacts.
In complex, multi-vendor or cloud-based environments, tracking isolated parameters within each component quickly becomes unmanageable. Even in a seemingly streamlined, all-in-one contact center, the reality is far from simple. Beneath the surface lie agent desktops, headsets, browsers, CRM and backend integrations, and network connectivity—which can vary significantly depending on whether agents are in-office, remote, or outsourced. And that’s before introducing the added complexity of a fragmented, multi-vendor stack. If each of the individual technology solutions is a “small box,” then the entire cloud solution is a “big box” that contains the small boxes. With conventional monitoring tools, understanding what is going on in that “big box” involves taking the data from each of the “small boxes” and putting that information into a data lake to perform correlations.
By contrast, CX observability—which is an approach and a set of software tools—looks at the “outputs” (sometimes called signals) which occur within (and are, of course, produced by the individual technology solutions) the overall system. Some examples of outputs that can be collected by a CX observability solution include audio cues (silence, mute states, hold music), call quality metrics (jitter, latency, packet loss), and interaction metadata (queue, agent, Caller ID).
A CX observability tool analyzes those outputs and uses machine learning / artificial intelligence (ML/AI) techniques to find patterns, correlations, and anomalies across interconnected elements. CX observability is a way to deal with that complexity by finding patterns in the outputs—e.g., what is working, what is not —and then flagging those issues so that a human can more effectively and more quickly troubleshoot the root issue(s).
The concept of observability was first applied to cloud infrastructure by AppDynamics (now part of Cisco), Datadog, Dynatrace, and New Relic, and subsequently became essential for managing complex cloud-based environments. This approach is now being brought to managing complex CX systems. The following provides a few examples of how emerging CX observability tools can be used not just to troubleshoot issues, but to drive continuous performance improvement.
Observability is a powerful end-to-end troubleshooting tool, capable of analyzing conversations across all supporting elements—applications, networks, CPU usage, headsets, and more. In short, it helps troubleshoot issues that are hard to find based on the individual monitoring of each element within a broad implementation.
Take, for example, a large manufacturing company with over 15,000 employees across multiple countries that used observability to diagnose intermittent audio issues affecting its distributed workforce. Agents in small offices relied on different network connections, making the root cause of those issues difficult to pinpoint.
The legacy approach to troubleshooting would have involved routine test calls to assess end-to-end connection quality. But, by using observability, the problem was diagnosed as the product of a combination of network and headset issues—two separate, randomly occurring factors that traditional tools would have struggled to identify. In this case, obsercability isolated the issues, enabling the company to resolve its infrastructure challenges efficiently.
Additionally, a CX observability tool can incorporate agent feedback through a help button—e.g., a lightweight extension running on the agent’s desktop—which would allow the agent to report the issue in real-time.
A CX observability tool creates an intelligent graph that connects all collected data points. By using AI, the observability tool can uncover patterns across that data graph.
A large bank leveraged CX observability to identify call avoidance behaviors, revealing trends such as agents frequently switching their states to game the system or taking an excessive number of breaks. These patterns can be programmed into the observability tool so that when they occur, supervisors are alerted. Similarly, the observability tool can track and flag excessive hold and mute times, the overall number of call transfers, and call durations that are too long or too short. Any of these “outputs” can indicate CX risks and thus the observability tool will alert a human supervisor who can investigate and take action to address the issue.
Note that observability tools can be applied to both human and AI agents. For example, the tools can uncover redal-time audio issues (e.g., latency and jitter) that may arise with either human or AI agents. CX observability tools can also because it examines the overall experience versus looking at the self-service and the human service as two disconnected experiences. CX observability can also uncover “doom loops” in which a customer in the self-service system is never able to reach a human agent. And, it can provide the analytics needed to strike the right balance between automation and human service by detecting when certain types of calls, which are first directed to self-service, have a high rate of escalation to a human.
As mentioned, observability can reveal patterns in data, and it shines when connecting different types of data. A large healthcare payer was facing quality issues with its government-sponsored programs, which were affecting Centers for Medicare & Medicaid Services (CMS) quality reporting and having a negative impact on revenue. The problem was compounded by having five Business Process Outsourcers (BPOs).
Because there were different types of data involved across multiple BPOs, legacy monitoring tools could not identify the root issue. So, the healthcare payer used observability to understand the context of transfers and distinguish normal ones from those signaling a problem—e.g., transfers occurring within the first few seconds of a conversation, transfers back to the same queue, or correlations with agent location, whether in the office or working from home.
CX observability hinges on modern data architecture to handle and correlate massive numbers of data points (i.e., outputs). The number of data points per interaction rapidly reaches 20 to 30 and can exceed that. This scale is what enables observability to reveal patterns and insights that traditional monitoring cannot. By layering AI and leveraging a robust data architecture, CX observability connects disparate data points across technical, operational, and experience layers. This capability enables organizations to track trends at scale, uncover root causes, and programmatically detect patterns that impact service delivery.
The data collected goes beyond technical information (e.g., call quality metrics) and includes operational data such as call duration, whether the customer was put on hold and for how long, transfers, disconnects, and more. The concept of uncovering patterns in a data graph that connects all data points can reveal insights not visible through traditional reporting. It offers a new perspective on improving the experience by delving into the "why" behind the data.
Consider hold time, for example. Every organization aims to reduce both the frequency and duration of holds. But customers may be placed on hold for various reasons: checking a system, which might indicate a system issue; consulting with colleagues, which might point to process or routing issues; or, if the hold occurs at the end of a call, agent behavioral issues might instead be the reason. It is only by using the observability tool and AI capabilities that hold times can be correlated with other events, and either benign or less than ideal patterns can be identified and, if necessary, action taken.
An increasing number of cloud contact center providers, including Genesys and NICE, are incorporating observability capabilities into their services. Meanwhile, CX assurance providers like Cyara and Nectar are expanding their testing and monitoring capabilities, while pure-play vendors focused on CX observability, such as Operata and Virsae have entered the market.
Some innovators in the industry are pushing the envelope by leveraging their data graph to combine technical, operational, and experience data. For example, Operata has integrated the data sourced through its monitoring with Amazon Connect Contact Lens and Amazon Contact Trace Records (CTR) detail records. Operata’s observability product connects all transactional details from the interaction, including call transcription and insights from unstructured data, such as sentiment, in its data graph. This creates a dataset that blends technical data, such as audio levels, operational data, such as call duration or time spent on hold, and experience data, such as sentiment or interaction outcomes. The combination of these different data elements allows Operata’s product to uncover new correlations, discover patterns, and program them for tracking.
Observability has become essential for managing complex network, infrastructure, and cloud-based environments—not just to troubleshoot issues, but to drive continuous performance improvement. CX applications and cloud contact centers connect customers, agents, and applications across complex, distributed systems. CX observability solutions can connect and correlate technical, operational, and experience data, thus helping organizations to maintain performance, detect issues proactively, and deliver uninterrupted experiences.
Nicolas De Kouchkovsky is the founder and principal of CaCube Consulting, an advisory and consulting firm helping B2B software companies grow.