You'd be hard pushed right now to find a board or a senior executive team that isn't considering leveraging AI to help them reduce operational costs, improve productivity, and boost efficiency.
As a result, many operational leaders and executives are under pressure to deliver on those board-level and C-suite aspirations.
Nowhere is this more true than in the service and support space, where the cost of labor has always played an outsized role in operating budgets compared with other departments and functions.
For example, research shows that headcount/labor, on average, accounts for approximately 70% of all operating costs of a service organization, while labor costs across other business functions often range from 15% to 30%.
Therefore, it should come as no surprise that service and support leaders are under pressure to reduce labor costs, with research by Gartner showing that 79% of service and support leaders expect to operate with a lower headcount in 18 months' time, primarily through increased use of AI.
But what if those expectations were not realistic or accurate?
I don't mean expectations around efficiency or productivity. It's clear that the right AI tool or system, implemented effectively and fueled by the right data, will, if pointed at the right problem, in all likelihood deliver efficiency and productivity benefits.
That's not the bit I am questioning.
I'm talking about cost-saving expectations and the implicit assumption from boards and executive teams that goes something along the lines of: 'We want you to do way more for way less.'
I say this for two reasons:
One, with the advent of AI, the nature of enterprise software has fundamentally changed. Historically, maintenance and development of a software platform largely sat with the vendor. Now, with the advent of AI-powered systems, that responsibility is shared with buyers, who are increasingly responsible for the customization, development, and management of their own instance.
This is having an impact, particularly in the service and support space, on the types of people and skills that service and support leaders need to train or hire to support and develop their AI platforms.
And, two, while we are still relatively early in this new AI-powered era, I don't believe we have yet achieved a clear picture of the true economics of AI operations that captures the costs of compute, inference, licensing or usage.
The Hidden Cost Equation
My concerns are supported by a recent Gartner research report, The State of the Human-AI Workforce in Service and Support, which finds that as automation levels approach 50% across a service and support estate, additional licensing costs, increased expenditure on specialized labor to support AI implementation, management, and development, and usage costs may exceed the savings from any frontline headcount reductions.
As the Gartner report notes:
