Keith Dawson's Analyst Perspectives

With a flood of artificial intelligence-related tools now available for contact center buyers, it can be helpful to take a step back and review where buyers can expect a quick return on investment and maximum improvement in efficiency and productivity. For now, it appears that there are three broad buckets into which contact center-focused AI applications are clustered. They all relate to agent behavior and activity, including:

• Front end containment and deflection via self-service, meaning the chatbot/virtual agent universe.
• Back-end applications that use AI to produce or analyze information that comes from or is delivered to agents.
• And apps related to measuring quality, coaching, training and other in-center actions.

We find a lot of variation and a bit of overlap within each of these three buckets. What's interesting to me is that most of the applications take existing processes or actions that agents deal with and then either add or subtract some motion to improve productivity.

For example, chatbots and virtual assistants that raise the containment rate affect the content and quantity of interactions that ultimately reach agents, changing the nature of the skills they need and the key performance indicators that measure success. By siphoning off the bulk of rote, procedural interactions, automated AI systems leave behind longer interactions that often have higher stakes for the customer relationship or higher value transactions that require a human touch.

On the back end, we find applications that use enterprise knowledge resources to deliver guidance during interactions and analytics tools that suss out customer sentiment and buying intent. The former moves the needle on standard contact center KPIs like average handle time and customer satisfaction. The latter applications do more for sales and marketing teams that use the information to orchestrate processes leading to desired outcomes—many of which relate to revenue and company health.

Finally, the third group is very clearly aimed at reducing the time-costing burden of after-call work, supervisor evaluation, shift trading and agent self-management. New applications for GenAI are emerging all the time: Take any aspect of the agent/supervisor work model and you can find processes that are inherently inefficient, repetitive and time-consuming, and therefore costly and degrading to the customer experience. The vast majority of these processes can—and are—being automated through well-tailored AI applications coming from many providers in the marketplace.

When we look at the way AI has been rolled out in contact center tools and platforms, we see an increasing focus on identifying the immediate benefits and return on investment. This has been especially evident in transcription/summarization, compliance monitoring and raising containment rates. Applications with a clear return on investment appear to be the most popular in these earliest days of AI rollouts.

But (and there’s always a “but”), the contact center has always been a conservative entity in its approach to technology innovation. It took the better part of a decade to adopt computer telephony integration in the ‘90s. That was a no-brainer, too, but it was significantly disruptive to make it happen. It took about five years to move from time division multiplexing to internet protocol at the turn of the century. It’s taken almost 20 years to get from on-premises to cloud, and we’re still not even halfway to a full conversion.

The lesson is that the bulk of enterprises will be slow to invest in AI for contact centers without:

• A clear demonstration of immediate cost-reduction
• A pathway to partial or sequential deployment of different apps in different contexts over time
• Clear examples that identify peer companies across industries with demonstrated success
• A genuine assessment comparing what the buyer wants to accomplish to the disruption in operations that deployment might cause, and
• A pitch that includes buyers outside of the traditional contact center and IT groups, especially since marketing and sales are the analytics consumers of the AI learnings about customer behavior.

All told, I think contact center buyers want assurance that what’s being created is essential to cost control right now and that it will also provide operational flexibility for the foreseeable future. ISG Research believes that by 2027, customer experiences will be largely driven by processes controlled by automated GenAI tools.

One of the most important lessons learned in the past two years is that buyers are genuinely interested in AI applications that demonstrate quick time to value. In other words, show a contact center a tool that automates an agented task or speeds a case to resolution, and you’ll be able to have an extended conversation about how that tool can be integrated into existing technology. On the other hand, start the discussion from the top down, arguing that AI should be everywhere, pervasive across applications in a platform, and the conversation is trickier and likely longer as buyers ponder the potential for disruption and the downsides of transformation.

It also suggests that there is common ground between the needs of various teams that weigh in on AI deployments. When decision-makers express a desire for “value optimization,” that’s akin to how contact center managers strive for cost control and productivity improvements. In a recent study (ISG Market Lens, 2024 AI Study, n=200), ISG Research asked enterprise decision-makers to identify specific AI applications currently in deployment; 64% cited customer chat and 57% cited predictive analytics/forecasting.

What I take away from these points is that early-stage conversations with customers must emphasize practical, immediate results from specific applications. I believe that after this first wave of successful applications has shown a year or two of real gains, conditions will be ripe for more expansive discussions that take AI from narrow approaches to broader options, making more use of AI as it relates to interdepartmental and back-office workflows.

Today’s value-realization proof points grow more apparent every day, and enterprises are increasingly alert to the opportunities. The next step is to use these successful use cases to build consensus among contact center leadership, the IT teams that work with them and executive teams, all based on a shared understanding of how AI creates specific, measurable results in productivity and efficiency.

Regards,

Keith Dawson