Contact center data is one of the most underutilized assets within modern organizations, despite offering a direct window into customer behavior, sentiment, and unmet need. Every interaction, whether a complaint, query, or cancellation, contains rich, unfiltered insight into why customers make decisions. Yet in many businesses, this data remains siloed, inconsistently captured, or reduced to surface-level metrics like call duration, abandonment rates, or resolution time, metrics that show what is happening but rarely explain why it is happening or where operational improvements should be prioritized.
Using artificial intelligence-assisted analysis and expert interpretation to transform this data into structured operational insight is a vital step in scaling automation successfully, surfacing not only what customers are saying but identifying where journeys are best suited for process redesign, self-service, or automation to deliver measurable operational improvements. The result is a contact center that becomes not just a service function but a strategic source of insight for improving customer journeys, reducing cost to serve, and guiding AI investment with confidence.
Contact centers have been quick to grasp the opportunities offered by AI and automation in theory. From early chatbot deployments, there has been a rapid push toward AI-driven orchestration and automation, with the goal of freeing up human expertise to deal with the most complex interactions. And yet, far too many contact centers have failed to achieve the desired goals. Of course, automation and AI can help reduce abandonment rates and improve resolution time, but only if organizations understand the underpinning causes of abandonment and slow resolution. Simply layering another technology on top of existing flawed processes and friction-filled customer journeys will never meet expectations.
The current lack of clarity and understanding regarding how to automate effectively and successfully is due, in part, to the focus on technology before understanding what is required and where it is needed. The temptation to explore the native AI now embedded in every application is only adding to the confusion. When a contact center is using CRM, contact center-as-a-service , workforce engagement management, and customer service management systems, deploying disconnected AI capabilities across each platform often creates additional operational complexity rather than measurable value. Furthermore, as these models are rarely optimized for the highly specific nature of contact center interactions—typically short, question-and-answer responses—or the vertical market-specific requirements, they lack the accuracy required in any customer-facing situation.
Yet companies already own the resources required to provide clear insight into innovation strategies. Contact centers generate vast volumes of interaction data at scale: information within these diverse operational systems or customer interactions via voice, text, chat, and email. This is an extraordinary resource that can shed light on every aspect of the contact center experience if it is correctly analyzed and reviewed.>
Turning Data into Decisions
Using AI-assisted analysis and expert interpretation, this data reveals the underlying drivers of demand, friction, and operational inefficiency. Consolidating these diverse operational data sources into a single, structured view reveals so much more than surface-level abandonment metrics. It highlights the root causes of repeat customer contacts. It identifies where customer journeys are breaking down and exposes gaps between brand promise and actual experience. And it also identifies repeatable, high-volume interactions suitable for automation.
Using this rich contact center data to ensure automation initiatives are evidence-based and optimally aligned with actual operational needs completely changes the focus.& It might reveal, for example, that AI automation is not the priority: If the customer journey is flawed and can be changed by simple process redesign, there will be an immediate improvement in contact center outcomes.
Indeed, this detailed data-led insight is key to understanding how and where AI can best be deployed. Identification and verification (ID&V) is a prime example. According to recent research, 74 percent of inbound calls to a contact center require ID&V to progress to the next step. And 91 percent of those calls are still handled by a human agent. So, while the actual customer request to access their records or check test results could be handled by AI, the need for a human to undertake the initial ID&V process removes an opportunity for automation: A human-to-bot handoff would break up the journey and create unnecessary friction for the customer. Automating ID&V removes one of the biggest operational barriers to scaling self-service and AI within the contact center.
Leveraging SLMs Smartly
Understanding this extraordinary data asset and what it reveals about the business is the key to unlocking value, whether the organization chooses to prioritize AI immediately or first address process and journey inefficiencies. And if AI is the route forward, it is important to consider which AI to use. Is it trained on contact center-specific interactions? Does it recognize the language used routinely within this vertical market? Areas such as healthcare and finance, for example, benefit hugely from small language models (SLM) that have been trained to recognize vertical market keywords allowing customers to be seamlessly routed to the correct AI or human agent.
SLMs can also be tuned using the behaviors and interaction patterns of high-performing contact center agents, mimicking their behaviors to further optimize outcomes. Building on this, micromodels can be deployed for highly deterministic, task-specific activities, such as ID&V that demands account numbers and vehicle registrations. Training a micromodel to understand alphanumerics is key to achieving the high levels of accuracy required. Plus, micromodels also require significantly less compute than frontier-scale LLMs, which is also a significant consideration for any business case.
Leveraging SLMs is also an important part of the continuous improvement process, improving relevance and accuracy to incrementally deliver more tailored insights from the data. SLMs also provide the context required for agentic AI to move beyond automating speech to automating actions. Combined with an automated ID&V model, agentic AI can manage the repeat actions: the test results and appointment creation that do not require human expertise.
The power of AI and automation to transform contact center performance and customer experience is compelling. It provides tangible opportunities to meet clear business goals, from reducing staff attrition to cutting costs and improving customer engagement. A successful strategy, however, demands understanding current problems, bottlenecks, and friction points.
The existing contact center data resource is the most valuable asset available. Using AI-assisted analysis and expert interpretation to create an optimization strategy, investment can be focused and measured. Organizations have the insight to prioritize high-impact changes, whether through process redesign, digital self-service, or targeted automation. The result is a contact center transformed from a cost center into a strategic engine for customer experience and operational efficiency.
Jack Godfrey is vice president of global sales at Connect.