AI Won’t Fix a Broken Customer Experience

I think most of us have had some version of this experience.

You have a relatively simple issue with your internet or mobile phone, so you contact one of the large Canadian telecom companies. You start online, because that seems like it should be the easiest option. You search for the answer, end up with a chatbot, explain what you need, and work your way through a series of questions that don't quite solve your problem.

Eventually, you decide you need to speak to someone. You authenticate yourself and get through to the call centre, re-explain the problem and discover you're speaking to the wrong department. You're transferred, wait again and explain the problem a third, maybe fourth time. Maybe the next person sends you somewhere else. Maybe you have to authenticate yourself again.

40 minutes later, your 2 minute problem is finally resolved.

What's interesting is that none of the individual pieces of that experience necessarily failed. The website worked. The chatbot worked. The call centre representatives may have been perfectly helpful and probably very nice. The problem eventually got solved. And yet the customer experience was terrible. Why?

They made it too hard.

That idea has always stuck with me from The Effortless Experience by Matthew Dixon. Companies spend enormous amounts of energy trying to delight customers when, much of the time, customers want something considerably simpler. They want things to be easy.

I think that's particularly relevant to the conversation we're having about AI.

There is tremendous excitement about what AI can do for Customer Organizations, and I share a lot of it. AI can handle customer interactions, identify risk, personalize journeys, summarize customer histories, improve self-service and potentially allow Customer Success teams to support far more customers. But I think we're sometimes starting with the wrong question.

Instead of asking “Where can we use AI?”, I'd start with: “Where are we making it unnecessarily difficult for our customers to get value?”

AI can't fix the underlying friction

Imagine our telecom company invests in an incredibly sophisticated AI agent. It understands what I'm asking, responds instantly and can resolve many issues without ever involving a person. That's a meaningful improvement. However, if I still end up being transferred between departments, repeating information I've already provided or navigating a process that was never designed around the customer, we've only improved part of the experience.

Customers don't care how we're organized internally. Those are our organizational realities, and they shouldn't become the customer's problem.

The same is true in SaaS. Every unnecessary handoff, repeated request or extra step creates friction between the customer and the value they're trying to achieve.

Before we ask how AI can improve the experience, we should make sure we've designed the right experience in the first place.

Start with the customer in mind

Before thinking about AI, I'd start somewhere much less exciting. Map the customer journey. Actually walk through what happens from the customer's perspective.

How quickly can they get started? When do they first experience meaningful value? Where do they typically get stuck? How many handoffs occur? What happens when they need help? What information do we repeatedly ask them for?

Then look for friction.

Maybe customers are entering information you already have. Maybe CSMs spend hours assembling information from five different systems before a meeting. Maybe customers contact Support for something they should easily be able to do themselves. Maybe every customer gets essentially the same onboarding experience regardless of what they're trying to accomplish.

Some of that friction will be obvious. Some of it has probably been around so long that nobody inside the company notices it anymore. Now the AI conversation becomes much more interesting.

We're no longer asking: What can AI do?

We're asking: Where can AI remove effort and make it easier for the customer to achieve value?

Build the foundation, then layer in AI

I've seen something similar happen with Customer Success technology. An organization buys a sophisticated CS platform expecting it to transform how it manages customers. Systems get connected. Dashboards get created. A health score gets designed. Automated journeys get built. Six months later, the organization has considerably more technology but isn't necessarily any better at managing customers.

The technology wasn't the problem. The organization hadn't made enough decisions with customer experience in mind.

AI carries the same risk.

If we haven't optimized the customer journey, AI can't do it all it for us. If we don't understand what drives adoption, AI isn't going to magically create that understanding. If our customer data is poor, giving AI access to more of it doesn't necessarily make the output better. And if a process is fundamentally frustrating for customers, automating it doesn't necessarily improve it.

That's why the foundation matters. Understand the journey. Get ownership right. Build sensible segmentation. Understand what value looks like. Create good self-service experiences. Decide where human interaction genuinely adds value.

Then layer AI into that foundation. That's when it gets exciting.

AI can recognize that a customer hasn't completed an important step and proactively guide them through it. It can spot declining adoption before the customer complains. It can instantly answer straightforward questions. It can summarize years of customer history before a CSM walks into a meeting. And it can remove hours of administrative work so people can spend more time helping customers solve meaningful problems.

Sometimes the best experience is almost invisible

Go back to the telecom example. Imagine instead that the system already knows who I am. It understands my account and my history. It recognizes the issue and, where possible, fixes it without me needing to contact anyone. If I do need help, I'm immediately directed to the right place. And if I need a person, that person already knows what happened, what I've tried and what needs to happen next.

I don't repeat myself. I don't navigate the organizational chart. It just gets solved and the customer experience is infinitely better.

AI may play an enormous role in making that possible. But the difference is that we're using AI to enhance a well-designed experience rather than compensate for a poorly designed one.

And that doesn't mean eliminating people. Sometimes the easiest experience involves a person. Sometimes it doesn't. If I can find an answer myself in 20 seconds, making me schedule a meeting with a CSM isn't high-touch service. It's friction.

But if I'm struggling to drive adoption across a complicated organization, I probably don't need another automated email. I may need an experienced person who understands my objectives and can help me work through the problem. The opportunity is to get much better at knowing the difference.

Start with effortless. Then make it intelligent.

I'm very optimistic about what AI will mean for Customer Organizations. But I wouldn't start with AI. I'd start with the customer.

Where are they struggling? Where are we making them work harder than they should? Where have our organizational structures become their problem? Where can we remove a step, eliminate a handoff, simplify a process or let the customer accomplish something themselves?

Build that foundation first. Then use AI to make the experience faster, smarter, more proactive, more personalized and more scalable. Because customers aren't buying our Customer Success model, our organizational structure, our technology stack or our AI strategy.

They're buying an outcome. They are solving business challenges. Our job is to make it as easy as possible for them to get there.