Reliance on AI tools in decision-making can lead to risky CX business

Dive Brief

Nearly 2 in 5 leaders said they experienced unintended CX-related consequences from using AI-powered tools in their decision-making process, per a TheyDo survey of 1,000 business decision-makers released last week.
Of those experiencing CX issues, approximately half said they saw increased customer complaints or support queries, and nearly half saw confusing or inconsistent customer experiences.
TheyDo’s research doesn’t assert that AI is causing these issues. It suggests a pattern: AI assistance can help organizations make faster decisions, but its suggestions may be based on data that conflicts across systems and teams. “A customer journey crosses far more boundaries than any single dashboard, system or department can see. A decision that looks sensible within one function can create friction somewhere else in the journey,” Jochem van der Veer, co-founder and CEO at TheyDo, told CX Dive in an email.

Dive Insight

As customer journey data resides in multiple back-end systems, decisions based on any single data point, or even several, can create unintended friction points.

“If the underlying data is incomplete, outdated or biased, the recommendation will be, too,” Julie Geller, principal research director at Info-Tech Research Group, told CX Dive in an email.

TheyDo’s research contends that creating context from those discrete data points is what matters. AI-powered tools can provide context, assuming integrations are made into those systems and the data resident in them is kept fresh and accurate.

Yet AI-generated CX recommendations from that context isn’t necessarily enough. CX leaders were 3.5 times more likely to trust AI’s output when they could see the underlying customer journey behind the recommendation, TheyDo found.

“The real value is not having AI make the decision,” Geller said. “It is helping leaders see what is happening sooner, understand why it may be happening and decide where intervention will have the greatest impact.”

Nearly one-third of respondents said that CX problems first become visible during customer service interactions, the survey found.

“Service teams often see the consequences of upstream decisions before the rest of the organization does,” van der Veer said. “Those signals need to feed back into the decisions being made by product, marketing, operations and other teams.”

Even so, AI-powered summaries of, and insights into, customer feedback and behavior should not replace CX leaders’ efforts around speaking with, listening to and observing customers.

“AI engines aren’t perfect at distilling key themes from all the data,” Jon Picoult, founder and principal of Watermark Consulting, told CX Dive in an email. “In addition, a skilled CX practitioner — unlike an AI tool — can use qualitative research to probe customers and gain a deeper understanding of their needs, wants, hopes and fears.