As Chief Availability Officer (CAO), Lou Senko leads Q2’s hosting team delivering an enhanced customer experience.
getty
I was recently up late one night trying to fix my car with a fancy new tool that I couldn’t get to work properly and tried calling the company but their help desk was closed. So, I went on its website, where its support page bragged about a recent AI-empowered enhancement. Twenty-five search attempts later, I still didn’t have a solution to my problem. I did, however, find what looked like an internal company document comparing the features of my tool to a competitor’s tool, even describing the problem I was having. I logged onto the support system of the competitor and found its own document describing the problem and the workaround to fix it. I got my problem solved, but certainly not in the way the company that manufactured my tool intended.
My experience demonstrates the need to be smart when deploying AI tools for customer support because they have a way of accessing, synthesizing and sharing information that may have unintended consequences. That’s not to say that AI doesn’t have an exciting and vital role to play in improving the customer experience (CX).
Indeed, thanks to the widespread adoption of large language models (LLMs), we're far closer to realizing the ultimate goal of creating customers for life. It’s not just about one transaction or solving one problem. It’s about deepening customer relationships, creating partnerships and increasing customer lifetime value through intimate personalization.
It’s clear from recent polls that executives understand and want to leverage new AI tools for precisely this purpose. In a 2023 Gartner poll of 2,500 executive leaders, "38% of respondents consider customer experience/retention as their primary focus of generative AI investments...ahead of revenue growth (26%), cost optimization (17%) and business continuity (7%)."
One way companies have chosen to utilize the new tools is to give customers the power to solve their own problems without having to interact with a person. The theory is that customers prefer this model. (Of course, this offers better internal scaling, which helps the company’s bottom line.)
Not surprisingly, in a CMSWire survey from October 2023 of CX leaders, 60% expect AI to have a transformative or significant impact on their business. When asked where they expect AI and machine learning (ML) to have the greatest impact on their company's CX, "enabling customer self-service" led with 45%, followed by "gaining actionable customer insights" (44%), "improving customer retention" (34%), "helping us find customer experience problems faster" (32%) and "increasing customer lifetime value" (31%).
Many companies have already integrated and scaled internal use cases using AI to supercharge customer support teams, onboard new staff faster, close the gap between mediocre and stellar performers, and reduce the overall human effort. Because the tools are so effective in helping staff find solutions to problems, many companies want to turn these same tools outward and allow customers to interact with them directly. But it’s not that simple.
The tools are created and trained with a treasure trove of knowledge gleaned from playbooks, knowledge base articles and years of documented solutions within support cases. As customer service teams solve new problems, the solutions go into the training model, so the tools get progressively better at solving problems, ad infinitum.
However, because the systems are trained on knowledge from industry professionals, they rely heavily on technical jargon, and training customers to effectively query resources that use technical jargon is a nonstarter.
Enter LLMs. By incorporating them into the process, customers can ask their questions in natural language, which the LLM effectively translates to the jargon needed to locate the information to solve the problem. Unfortunately, this solution comes with a new set of challenges.
First, because the LLM returns a solution cobbled together from many different sources, it may lose its technical accuracy (e.g., the LLM finds the predictive next word from multiple unrelated documents that were never meant to be together). Because customers place a high degree of trust in the tool’s response, it’s important to ensure the answers are worthy of that trust.
Second, the answer may be so jargon-filled that the customer can’t do anything with it. Additionally, much of the knowledge needed to fix a technical problem is found in the private notes of customer service agents. Most support case systems allow companies to toggle what the customers are allowed to see and what's kept private, so they now need to allow customers to see those private notes to fix their problem, and those notes will need to be cleaned up for public consumption.
And then there’s the question of customer data security. Customer support teams need to be able to share fixes across their customer base in a way that doesn't unintentionally divulge private or proprietary information. Companies could allow the customer to search for answers but not return sensitive information to the person doing the search. It could also keep solutions for individual customers separate so one isn’t receiving private information about the other. Unfortunately, this separation of data could potentially reduce the effectiveness of the responses. If you tighten the scope to just the customers’ own data, they can’t leverage the larger collective wisdom to solve new problems.
These obstacles can be overcome, but they're complicated and require a larger investment of time and resources than many CX leaders plan for when implementing AI tools to enhance customer self-service. It’s also worth remembering that a vital part of creating customers for life is strengthening the personal connection, and this can only be done, well, in person. I can run through the self-checkout line myself or have a polite interaction with a person at the checkout counter and leave with a stronger, more pleasant memory connected to that business.
When customers have issues, we have opportunities to not just solve their problems but to deepen the customer relationship—one touch at a time.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
