You can’t upend the entire technological infrastructure for an industry without seeing some ramifications in the form of personnel shifting and dislocation. Contact centers are smack in the middle of a transformation that is remaking the toolset and forcing businesses to rethink both the fundamental purpose of their centers and the mix of managerial and operational employees.
Let’s consider how different roles within and around the contact center are going to change over the next three years. I think we start from the premise that the downstream effect of injecting artificial intelligence (AI) so heavily into contact center technology is that it reduces the need for customer service reps (CSRs) (I’m going to call them that to distinguish “agents” from AI agents).
What’s not clear is the speed with which centers will either go for reduced headcount dividends or will re-task agents into different functions.
We can make some starting assumptions about how things will play out, beginning with the fact that automation will take on the lion’s share of routine work. We already see this in how businesses handle first-tier customer contacts, basic after-call work and simple things like knowledge lookup. Generative AI (GenAI) has made it clear that time spent on interactions can be shrunk significantly by suggesting next best actions, summarizing interactions and other informational boosts.
We also know that humans are starting to be tasked with harder calls: those that require emotional responses, judgement calls, VIP customers. This shift was predicted years ago and is now coming true.
So, we’re expecting the CSR population will be reduced, especially among the minimally skilled/trained workers handling Tier 1 support. The question is: will those people be shifted into different roles or will their roles go away?
I think the industry would like to think they will be shifted, but that’s tricky for two reasons. One, you have to have skills, or training for skills, to shift from one job to another. There are investment considerations here because where people don’t have shiftable skills, they need to be upskilled. And two, work functions on a pyramid, and there are many fewer job opportunities in the scenario we’re about to explore. You just don’t need as many skilled workers in these other roles as you did to handle Tier 1 support at volume.
What kinds of shifts are we talking about?
For basic CSRs, we are seeing a couple of possible models. One is to create a sort of super-rep, trained specifically for those harder, more complex interactions, fully supported by AI and somewhat responsible for outcomes and for revenue (rather than for speed and volume).
The other model, similarly, is to view that super-rep as a lower-level supervisor, but what they’re supervising isn’t other people, it’s the bots themselves. They would handle the bot failures, the misroutes, the escalation queues. This requires some traditional CSR skills, but also a bit more process understanding—why did the bot fail, and what must I do to pick up the thread?
Move up the employment chain and you come to the supervisors. They will have, by definition, fewer people to manage and fewer process tasks like scheduling, queue management and manual quality evaluations. In this future scenario, they end up transitioning into more of a coaching role, where they’re also coaching these mini-managers that have been created out of the CSR pool on how to handle the processes and the bots, as well as customers.
Quality management and workforce management specialists will be needed for as long as enterprises are resistant to deploying AI in their operations. ISG Research asserts that by 2028, nearly all customer interactions will be screened by AI for quality and agent performance. In an aggressive deployment of AI, these folks will have to segue into more value-generating work in short order. If it takes some time for the entire industry to make changes, then I think you’ll still see some of these specialists in their current roles five years from now, or longer. Consider how slowly the contact centers have actually changed in the past, in contrast to how quickly software providers would like to see them change their tech stack.
This is contingent on the voice AI agent becoming the default mode of customer contact in the first instance. Industry interest in exploring AI agents is off the charts right now. But interest doesn’t equal deployment, and real production adoption in contact centers is still very early. If we ask how many enterprises are piloting programs that involve voice agents, the numbers are very high. There are a lot of tire kickers. But I think it’s reasonable to expect that by the end of 2027, we’re looking at about half of mid to large centers with at least one fully baked voice agent sitting in front of customers. Those centers that allocate a significant portion of their overall volume to bots will be lower, maybe half of that, or a quarter of the total.
What’s interesting about the shift to AI voice agents (however long it takes) is that a business can’t just flip a switch and pivot from one operating model to another. To make an effective transition, you need investment not just in the tools, but in process design, quality assessment, compliance and data protection, training people and training customers. That suggests a transition process that could take longer than people realize. If you doubt that, I’ll just say that one-third of all contact center seats are still on premises, and we’re 20 years into that switch.
So, the technology and workforce transitions will happen side by side for the foreseeable future, with many businesses building new roles on the fly as needs are exposed. We’re not sure what these roles will be called, but there are new functions that people can take on. For example, centers will need people to design “conversations” or interactions, which is a short step from being an orchestration designer of the overall journey.
Someone will be tasked with managing the knowledge sources that feed the automated customer-facing systems, a sort of content controller. This could be a significant role, as it can expand to incorporate marketing-related content, corporate knowledge management, product information management and potentially more sources. This role is closely related to the person tasked with training and evaluating the performance of the AI agents.
None of these roles is directly customer-facing, upending the entire skillset traditionally sought for contact center work: empathy and communications. And businesses will need fewer folks in these new roles than they have available employees in the center.
So how does this play out? Here is a back-of-the-envelope assessment. If today a typical center is maybe 70% populated by Tier 1 reps, another 10% supervisors/coaches and 10% in specialist Tier 2 support roles, then within about three years I think that becomes around 45% Tier 1, 15% supervisors/coaches and 20% specialists/Tier 2. That will occur against the backdrop of a baseline reduction of about 10-15% overall headcount.
Quibble with the figures all you like, but this is directionally where we’re headed. Given this, enterprise contact center leaders should plan for a sustained, multi-year transition in which automation and human labor coexist, but with structurally fewer entry-level roles and higher expectations for the remaining workforce. The practical implication is that workforce strategy must be treated as a core element of AI adoption, not a downstream HR problem.
To get ahead of the changes, start investing now in upskilling but be realistic: not all roles are shiftable, and new AI-adjacent roles will be fewer in number than displaced CSR positions. Organizations that acknowledge this early and design toward it will retain control over cost, quality and workforce morale; those that cling to optimistic assumptions about seamless reskilling or rapid, painless automation will be forced into reactive decisions later, under less favorable conditions.
Regards,
Keith Dawson
Fill out the form to continue reading.