Episode 249 | November 16, 2026

What service design knows about AI that UX teams don't

Explore how service design improves customer experiences, breaks down silos, and helps UX teams design human-centered AI agents with expert Stephen Taylor.

What service design knows about AI that UX teams don't

What service design can teach us about customer experience, organizational silos, and building AI agents that actually help people

Imagine taking time out of your lunch break to call your internet provider, only to spend the next 30 minutes fending off sales pitches when all you wanted was to transfer your service to a new address.

You're moving. You have a dozen other calls to make. And the person on the other end of the line seems determined to sell you something you don't need.

It's a frustrating customer experience. But here's the question: Is this really a customer service problem?

Not necessarily.

The problem might be hiding somewhere the customer never sees.

That's one of the central arguments design strategist Stephen Taylor makes in a recent episode of Insights Unlocked, where he joins host Jason Giles, UserTesting's vice president of customer intelligence, to discuss service design, customer experience strategy, and what these disciplines can teach us about designing AI agents.

After more than 30 years working across UX, research, and service design, Stephen has learned that improving an experience often means looking well beyond the experience itself.

And sometimes, the biggest opportunities have nothing to do with the interface.

The real problem was hiding in the call center

Consider that frustrated internet customer.

Stephen described a telecommunications company that was losing customers when they moved between addresses. The company wanted to understand why.

Researchers discovered that customers weren't necessarily unhappy with their internet service. They were unhappy with the process of transferring it.

When the team investigated further, they found something revealing.

Stream On

Share

Get Started Now

Contact Sales

BENCHMARK

AI Relationship Quality

The UserTesting AI Relationship Quality (ARQ™) benchmark gives you a clear measure of how well your AI experience is working for customers

Inside the call center, a massive screen displayed employee rankings for cross-sells and upsells. Individual agents also tracked their sales performance at their desks.

The incentives were clear. Sell more.

Unfortunately, those incentives were directly at odds with what customers needed.

"Their scorecard was optimized for this," Stephen explained, "and it was optimized precisely for what customers who were trying to move didn't want."

The solution wasn't another website redesign or a more intuitive customer portal.

The company created a dedicated call center for customers who were moving, with performance measures aligned to their needs.

Customer churn dropped 50%.

Play video

That's a remarkable business outcome from changing something most UX teams would never be authorized to touch.

And it illustrates a fundamental principle of service design: The experience customers see is often shaped by decisions they don't.

Service design means looking beyond the screen

Traditional UX design tends to focus on specific interactions: a website, mobile app, checkout flow, or digital product.

Service design takes a wider view.

It examines the entire customer journey, including the people, processes, policies, technologies, and organizational structures that make those interactions possible.

Think of UX design as renovating a room. Service design asks whether the plumbing, electrical system, and floor plan actually support how people want to live in the house.

Both matter. But fixing the paint won't help much if the pipes are leaking.

Stephen describes service design as shaping service delivery end to end, across channels and from the perspectives of everyone involved.

That includes customers, frontline employees, and the people working behind the scenes.

The distinction matters because organizations often optimize individual touchpoints while overlooking the connections between them.

A customer might have a terrific mobile experience but encounter an entirely different reality when calling customer support.

The interfaces work.

The service doesn't.

For UX and product teams, this suggests a practical question worth asking before launching another redesign:

Are we solving the customer's problem, or just improving the part of the problem we control?

Customer research can break down organizational silos

One of the more interesting ideas from the conversation concerns how service design teams collaborate.

Because service design problems often cross organizational boundaries, solving them requires people from different departments to develop a shared understanding of what's happening.

That sounds straightforward.

Insight+

Get on-demand access to some of UserTesting's most popular events, including from Crafted Seattle and Crafted London. Register for Insight+ and start watching today.

Anyone who has participated in a cross-functional planning meeting knows otherwise.

Engineering, marketing, product, operations, and customer support often approach the same problem with different priorities, incentives, and assumptions.

Stephen argues that customer research can help bridge those differences.

His teams bring stakeholders into research sessions, encourage them to observe customers directly, and involve them in interpreting what they've learned.

Rather than presenting research findings after the fact, they build understanding together.

"We tell stories and that becomes sort of the lingua franca of all our collaboration," Stephen said.

The approach also addresses something that dashboards and research reports often miss: the tacit knowledge people develop through direct experience.

Play video

Stephen draws on the concept of hermeneutics, the practice of interpretation, to explain why understanding customers requires more than collecting facts.

Two people can examine the same research findings and draw different conclusions. But when they participate in the research together, they have a better opportunity to understand not only what happened, but why.

For research leaders, there's a practical lesson here.

Don't just share customer insights with stakeholders. Invite them into the process of discovering those insights.

Alignment becomes easier when people have witnessed the same problems firsthand.

What service design can teach us about AI agents

The conversation takes an especially timely turn when Jason asks how service design principles apply to AI-powered customer experiences.

It's a question more product and UX teams should be asking.

As organizations introduce conversational AI agents, they're creating interactions that increasingly resemble exchanges between people rather than traditional software interfaces.

Stephen believes service designers have valuable experience to contribute.

After all, they've spent decades designing services delivered through human roles.

Those roles have responsibilities, goals, policies, permissions, and constraints.

AI agents need many of those same things.

"It's just that there isn't a human behind it," Stephen explained.

This is where service blueprinting becomes particularly useful.

A service blueprint maps the complete delivery of an experience, distinguishing between what customers encounter on the frontstage and the backstage systems, processes, and people supporting it.

For AI agent design, that means asking questions beyond whether the chatbot can successfully complete a task.

On-Demand Webinar

Evolve your UX research strategy to win in an AI world

Learn how leading UX teams combine AI and human expertise to accelerate research, reduce uncertainty, and make better product decisions.

What information does the agent need? What decisions should it be authorized to make? When should it transfer responsibility to a person? And what happens elsewhere in the organization when it takes action?

These aren't simply interface questions.

They're service design questions.

And they become increasingly important as AI agents move beyond answering questions to performing tasks on behalf of customers.

Not every interaction should be automated

There's another dimension to AI-powered customer experiences that deserves more attention.

Just because an interaction can be automated doesn't mean it should be.

Play video

Stephen points to his experience researching healthcare services, where patients frequently emphasized the importance of knowing someone genuinely cared about their well-being.

Sometimes, the presence of another person is part of the value being delivered.

Consider a patient navigating a frightening diagnosis.

An AI agent might efficiently schedule appointments, retrieve information, or explain administrative procedures.

But efficiency alone doesn't address every human need.

For teams designing AI-powered services, the challenge is deciding which interactions benefit from automation and which depend on meaningful human connection.

That's not an argument against AI.

It's an argument for designing the entire service rather than treating automation as the objective.

Start with the people, not the technology

Toward the end of the conversation, Stephen introduces another compelling idea: Organizations should pay closer attention to what makes employees find meaning in their work.

In healthcare research, he began asking nurses which activities frustrated them and which allowed them to do the work that originally inspired them to enter the profession.

The answers were sometimes surprising.

One nurse described being called into the hospital at 3 a.m. to care for a patient experiencing an emergency.

Stephen expected the experience to represent an exhausting burden.

Instead, the nurse found it deeply meaningful.

That was precisely the kind of work she had chosen nursing to do.

The lesson extends well beyond healthcare.

Organizations often spend considerable resources measuring performance and creating incentives while overlooking the intrinsic motivation employees already bring to their jobs.

Service design can help identify those moments when employee purpose and customer needs reinforce each other.

It also raises a useful question for AI implementation: Could automation remove the administrative work employees dislike, giving them more time for the human interactions they value?

That's a more promising starting point than simply asking how many tasks AI can replace.

For product, UX, and CX leaders, the larger takeaway is straightforward: Before redesigning an interface, deploying an AI agent, or optimizing another metric, examine the complete system delivering the experience.

Talk to customers. Observe employees. Question incentives. Map the service.

Then decide what actually needs to change.

As Stephen put it, "The giving is what they are getting back."

Additional resources

These resources are selected to complement Stephen Taylor's discussion of service design, customer research, cross-functional collaboration, and AI-powered customer experiences.

Frequently asked questions (FAQs) about service design and AI