AI in Customer Service: Fix the Experience First
AI Can't Fix a Bad Customer Experience. But It Can Make One Worse.
I think artificial intelligence is going to change the way businesses interact with their customers in some pretty remarkable ways.
Actually, it already is.
We're seeing AI-powered receptionists, virtual agents, intelligent call routing, automated customer support, and systems that can analyze conversations and provide information to employees in real time.
Some of this technology is genuinely impressive. And after nearly 30 years in the telecommunications and technology industry, I can tell you that the capabilities available today would have seemed almost unimaginable when I started my career.
But there's something about the current rush to implement AI that concerns me.
We're getting really good at automating customer interactions without always asking whether those interactions were any good to begin with.
And that's a problem.
The customer doesn't care how sophisticated your technology is.
Think about the last time you called a company because you needed help with something.
Maybe it was a billing issue, a service problem, or a question about an order.
You probably weren't particularly interested in the company's contact center technology. You weren't thinking about conversational AI, natural language processing, or how many customer interactions the company had automated.
You wanted your problem solved.
Preferably without spending twenty minutes explaining it to three different people.
Now imagine that same experience with an AI-powered virtual agent.
The system answers immediately. It sounds surprisingly human. It understands your question, asks a few follow-up questions, and assures you that it can help.
Then it transfers you to another department.
You explain everything again.
That department doesn't have the information it needs, so you're transferred somewhere else.
Eventually, you reach a person who tells you that your issue requires a different team.
Congratulations. The company has successfully automated the first ten minutes of a terrible customer experience.
And it probably has a dashboard showing how efficiently it handled your call.
That's the disconnect I worry about.
Automating a bad process doesn't make it a good process.
I've spent a significant part of my career working with businesses on communications, connectivity, and contact center solutions.
One thing I've learned is that technology problems aren't always technology problems.
Sometimes they're process problems.
A company might have an outdated phone system, but the real issue is that nobody has clearly defined who should handle certain customer requests.
Or perhaps the contact center has a sophisticated routing platform, but customer information is scattered across three different systems that don't communicate with each other.
Maybe employees are doing their best, but they don't have the authority or information necessary to resolve an issue during the first conversation.
None of those problems disappear just because you introduce artificial intelligence.
In fact, AI can make them more noticeable.
If a customer is being routed to the wrong department today, an AI-powered system might simply route them there faster.
If your CRM contains inaccurate information, an AI agent using that information can provide inaccurate answers with remarkable confidence.
And if your organization has never defined what a successful customer interaction should look like, AI isn't going to establish that definition for you.
At least not without some serious work on your part.
Cost savings are important. But what are we actually measuring?
I understand why businesses are excited about AI.
Labor costs continue to put pressure on organizations, customers expect faster responses, and finding and retaining good employees can be difficult.
The opportunity to handle routine questions automatically, operate around the clock, and give employees better tools is incredibly attractive.
And there are absolutely situations where AI delivers meaningful value.
But I think we need to be careful about how we define success.
If an AI implementation reduces the average cost per customer interaction by 30%, that sounds like a great result.
But what if the number of repeat calls increases?
What if more customers abandon their interactions because they can't reach the person they need?
What if customer satisfaction declines, even though the technology is working exactly as designed?
Have we actually improved the business?
Or have we just made it less expensive to disappoint people?
I would argue that first-contact resolution, customer satisfaction, repeat contacts, and the ability to reach a human being when necessary are at least as important as the cost of handling an interaction.
The numbers need to tell the whole story.
The best use of AI might be helping your employees.
Here's an area where I think AI has tremendous potential, and one that doesn't always get as much attention.
Instead of focusing exclusively on replacing human interactions, what if we focused more on making the people handling those interactions better at their jobs?
Imagine a customer service representative receiving a call and immediately having access to the customer's account history, recent service issues, previous conversations, and relevant troubleshooting information.
Imagine an AI assistant that can summarize a complicated interaction, suggest the next appropriate step, or locate the answer to a technical question while the employee is still speaking with the customer.
The employee can focus on the conversation instead of searching through multiple applications.
The customer doesn't have to repeat information that the company should already know.
And the business gets a more consistent experience without sacrificing the personal interaction that many customers still value.
That's a pretty compelling use of technology.
I'm not suggesting every customer interaction needs a human being. Routine requests like checking an order status, scheduling an appointment, or answering a straightforward question are often excellent candidates for automation.
But there's a difference between giving customers a convenient self-service option and making it nearly impossible for them to speak with someone when they genuinely need help.
We shouldn't confuse the two.
AI needs a foundation, just like every other technology.
There's another part of this conversation that tends to get overlooked.
AI doesn't operate in isolation.
It depends on the systems, data, networks, and processes supporting it.
If you're deploying AI in a contact center, you need to think about the reliability of your connectivity, the quality of your integrations, the accuracy of your customer data, and the security of the information being exchanged.
You also need to consider what happens when something goes wrong.
- What does the customer experience look like when the AI can't answer a question?
- How does the interaction move to a human employee?
- Does that employee receive the conversation history, or does the customer start over?
- Who is responsible for monitoring the quality of the responses?
These aren't particularly glamorous questions, but they're the ones that often determine whether an implementation succeeds.
And they're exactly the kinds of questions a good Technology Advisor should be asking before recommending a solution.
Before buying AI, fix what frustrates your customers.
At Utel USA, we're having more conversations about AI and automation as businesses look for ways to improve efficiency and customer engagement.
I think that's a positive development.
But I also believe we have a responsibility to challenge the assumption that adding new technology automatically creates a better outcome.
Before a company invests in an AI-powered customer experience platform, I'd encourage its leadership team to spend some time examining the experience customers have today.
- Where do they get frustrated?
- Why do they call back?
- What information do employees struggle to find?
- Which problems could be eliminated entirely with a better process?
- And which interactions genuinely benefit from having a person involved?
Those answers should influence the technology decision, not the other way around.
I'm excited about what AI can do for businesses. I think we're only beginning to understand its potential.
But I've been around long enough to know that impressive technology and good business outcomes aren't always the same thing.
If your customers already hate your phone system, giving it artificial intelligence probably isn't going to change their minds.
Maybe we should fix the experience first.
Then let AI help us make it even better.











