The Function That Spent the Most and Proved the Least

Service and support put a larger share of budget into artificial intelligence than any other business function, and three quarters cannot show a return.

Sign up for my Substack daily AI newsletter here.

See my AI Training course portfolio for corporate Business Leaders here.

Article link: https://johanosteyn.substack.com/p/the-function-that-spent-the-most?r=73gqa&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true

Follow me on LinkedIn: https://www.linkedin.com/in/johanosteyn/

Of ten business functions assessed in recent research, the one that committed the largest share of its budget to artificial intelligence was customer service and support. It is also the function least able to demonstrate that the money worked. That combination should interest any board that approved the spending, because the explanation turns out to have very little to do with the technology.

CONTEXT AND BACKGROUND

A survey of 1,303 senior leaders across industries, conducted between January and April 2026, found that service and support leaders invested a median of twelve per cent of their 2025 budget in artificial intelligence, the highest proportion among the ten functions measured. Only twenty four per cent of those leaders could demonstrate positive financial returns across their use cases. Set that against what the market was promised. Analysis published in February projected that generative artificial intelligence could reduce the cost base by twenty to thirty per cent in customer service within two to three years, alongside enhanced customer satisfaction and quicker resolution. The projection may still prove right in time. The measured result so far is that three quarters of the leaders who spent the most cannot evidence a gain.

INSIGHT AND ANALYSIS

The diagnosis offered by the analyst who led the research is worth sitting with, because it exonerates the technology and implicates the design. The disappointing impact of customer-facing investment has less to do with technology limitations than with misalignment with customer expectations. Customer behaviour has shifted underneath the deployments. In a separate survey of 3,566 business and consumer customers conducted in February and March 2026, use of third-party generative tools during service interactions was found to have nearly doubled in a year, while use of company-provided chatbots has remained statistically unchanged since 2022. Customers are now roughly three times more likely to reach for an outside model than for the one their supplier built, and fifty eight per cent of users have had generative artificial intelligence complete a task on their behalf, rising to seventy four per cent in business-to-business settings. Most corporate chatbots were built to answer questions. Customers now expect them to act.

IMPLICATIONS

Two consequences follow, and neither is a technology problem. The first is measurement. Contact centres continue to be judged on average handle time, cost per contact and call deflection, while customers judge them on whether the problem was solved. After repeated negative contact centre experiences, twenty eight per cent of bank customers reduced their spending and thirty one per cent stopped doing business with the institution altogether, and forty per cent of the executives surveyed did not rank improving customer experience among their top three priorities for contact centre strategy. The second is candour about capability. Research on workforce decisions found that many organisations announcing artificial intelligence as the reason for change had no mature, vetted application ready to do the work, a pattern the researchers named directly as attributing financially motivated decisions to future capability.

CLOSING TAKEAWAY

The corrective is cheaper than the original investment. Customers are not rejecting the technology. Fifty per cent say their interactions are easier when companies use it, while eighty seven per cent say the option to reach a human being is essential, and when customers unwilling to engage were asked what would change their minds, the most common answer was the ability to switch to a person. Generative artificial intelligence should therefore not be a mandatory first step for every issue, because customers pushed through repeated unsuccessful attempts before reaching help stop using the tool at all. The question for the board is narrow and answerable. What did the service function spend last year, what can it evidence in return, and did anyone ask whether the system was designed around what customers were actually trying to do?

Johan Steyn is a prominent AI thought leader, speaker, and author with a deep understanding of artificial intelligence’s impact on business and society. He is passionate about ethical AI development and its role in shaping a better future. Find out more about Johan’s work at https://www.aiforbusiness.net