An exploration for strategy leaders, enterprise architects and customer experience professionals
Something profound is happening at the intersection of artificial intelligence and customer experience — and most organisations haven't fully caught up with it yet. We talk a lot about AI changing how we work, how we code, and how we communicate. But what about how we serve customers? Specifically, how we think about what it even means to give a customer a good experience online?
This post is about a fundamental shift happening right now in the IT industry — a shift in the mental model that underpins how we design, build and measure digital customer experiences. Whether you work in enterprise architecture, digital strategy or customer experience, this will be relevant to you. Let's get into it.
The World We Built: Traditional Website Success Metrics
For the better part of two decades, the success of a service provider's digital presence — be it a mobile network, a broadband company, a utility, or your local council — was measured through a well-established set of KPIs. These metrics were sensible, data-driven, and shaped how entire product and engineering teams prioritised their work:
- Session duration / time on site — how long visitors spent browsing
- Pages per session — how many pages a user clicked through in one visit
- Bounce rate — the percentage of visitors who left after viewing just one page
- Unique visitors and traffic volume — the sheer reach of your digital front door
- Conversion rate — how many visitors completed a desired action (purchase, sign-up, enquiry)
- Organic search traffic and keyword rankings — your visibility on Google
- New vs. returning visitors — measuring loyalty and campaign effectiveness
- Page load time — technical performance and user retention
- Customer Lifetime Value (CLV) — long-term revenue from an ongoing customer relationship
These metrics shaped how websites were built. If you wanted customers to view their bill, you built a billing portal with multiple pages. If you wanted them to browse handsets, you built a catalogue. If you wanted to upsell a broadband upgrade, you built a multi-step journey with plenty of touchpoints. The longer a user stayed and the more pages they clicked through, the better. Engagement meant value. Time on site meant interest. Pages per session meant you were guiding them well.
This logic wasn't wrong — it was appropriate for the era. And it produced genuinely useful products. Self-service portals replaced call centres. Online ordering reduced friction. Customers gained more control over their accounts and services. It was progress.
But here's the uncomfortable truth: that progress was built on an assumption that no longer holds.
The Assumption That Broke
The assumption was this: customers come to your website to browse.
That assumption drove everything. It explained why broadband providers built elaborate comparison tools. Why mobile operators created beautiful handset galleries. Why councils designed step-by-step service directories. The design philosophy was: bring people in, walk them through, help them discover what they want.
But increasingly, that isn't why people visit websites. Customers today — particularly in service-heavy industries like telecoms, utilities, and public services — arrive with a very specific intent already formed. They have already decided. They don't want to browse. They want to act. And every extra page, every navigation menu, every click between them and their goal is friction. Frustrating, trust-eroding, relationship-damaging friction.
Think about how you personally use a service provider's website today. Do you browse? Or do you arrive knowing exactly what you want to do — check why your bill has gone up, report a fault, upgrade your plan, find out when a technician is coming — and just want it done?
If we're honest, it's almost always the latter. And AI is about to make this the overwhelming norm.
The AI-Powered Shift: From Navigation to Resolution
Now imagine a different experience. A customer lands on your homepage and, instead of a navigation bar with ten dropdown menus, there is a simple prompt field. One box. And they type:
In the traditional model, each of these requests would require the customer to navigate through several pages, log into a portal, click through menus, maybe wait for a call centre. In the AI-powered model, the response is immediate, personalised, and complete. The customer's purpose is fulfilled in seconds.
This is not science fiction. The industry data tells the story clearly. A 2024 survey by Bain & Company found that around 80% of consumers rely on zero-click results in at least 40% of their searches — meaning they expect their answer to surface instantly, without further navigation. Meanwhile, the number of AI assistant prompts grew by nearly 70% in the first half of 2025 alone, as customers increasingly bypass traditional search and browsing entirely.
The KPMG Global Customer Experience Excellence report for 2024–2025 — based on over 86,000 consumer interviews across 23 countries — found that personalisation has become the single biggest driver of customer loyalty. Not price. Not product. Personalisation. And personalisation, at scale, requires AI.
Two Different Mental Models
Let me put the contrast as clearly as I can, because I think this is the crux of the strategic challenge:
The traditional approach asks: "The customer wants to view their bill. How do we make that possible?"
The AI-powered approach asks: "Why does the customer want to view their bill? And how do we resolve that underlying need before they even have to navigate anywhere?"
A customer who checks their bill usually has one of a small number of reasons: they think it's too high, they want to dispute a charge, they're wondering if they're on the right plan, or they need to share the information with someone else. Each of these is a distinct need, and each has a distinct resolution. The traditional model treats them all the same — here's a PDF of your bill, good luck. The AI model identifies which scenario applies and responds accordingly.
This is what I mean when I say the shift is about intent, not access. We've spent years making information accessible. The next era is about understanding intent and resolving it.
What This Means for Enterprise Architecture
This shift has significant implications for how enterprise systems are designed. The traditional modular approach — build a billing feature, build a service catalogue, build an order management module, integrate them — made sense when you were building pages. The mental model was: one feature, one page, one journey.
In an AI-driven model, the architecture needs to think differently. The AI layer needs to reach across all those modules simultaneously. When a customer asks why their bill is higher, the AI must be able to access billing history, usage data, promotional expiry dates, and price change records — all at once — and synthesise an answer. That requires a fundamentally different approach to data integration, APIs, and system design.
Bain & Company's recent research makes this point directly: capturing the full potential of agentic AI will require modernising enterprise architecture. This is not a bolt-on. It is a re-architecture challenge that requires leadership alignment between digital, IT, and customer experience teams.
BCG's 2024 research adds an important warning: only 26% of companies have built the capabilities needed to move beyond AI proof-of-concept and generate tangible value. The primary obstacles are not technical. They are organisational — people and process, not algorithms. This means that for most enterprises, the hardest part of this transformation is not deploying an AI model. It is redesigning the workflows and governance structures that surround it.
What About Upselling? The Commercial Question
I can hear the pushback already: "But our traditional multi-page journeys create opportunities to upsell. If we resolve the customer's query instantly, we lose those touchpoints."
This is a legitimate commercial concern — but it is also a false dilemma.
Consider how a well-designed AI interaction can handle this. A customer asks why their bill is higher. The AI tells them: their promotional discount expired last month, which added £8 per month to their plan. It then immediately offers: "I can check whether you're eligible for our new loyalty pricing, which could save you £5 per month — shall I look into that for you?"
That is a far more effective upsell than burying a promotional banner on page seven of a billing journey. It is contextually relevant, it arrives at the precise moment of customer attention, and it solves an actual problem the customer has.
Early adopters of AI-powered customer experience are already seeing this play out. Research from Metrigy's AI for Business Success study shows companies implementing AI in customer experience are seeing an average 26.7% lift in revenue and a 32.6% improvement in customer satisfaction scores. That is not the story of a trade-off between efficiency and commercial performance. It is the story of both improving simultaneously, because relevance and speed are what customers reward.
The New Metrics: What Should We Measure Instead?
If we accept that the old metrics are no longer fit for purpose, we need to be clear about what replaces them. Here is a suggested framework for AI-era customer experience measurement:
- Task completion rate — did the customer achieve what they came to do?
- Time to resolution — how quickly was their intent fulfilled?
- First-contact resolution rate — did they have to come back, call, or escalate?
- Intent recognition accuracy — is the AI correctly identifying what the customer actually needs?
- Containment rate — what proportion of interactions are resolved fully by AI without requiring human escalation?
- Customer effort score — how much work did the customer have to do?
Notice what is absent from this list. Time on site. Pages per session. Bounce rate. Those are not the metrics of a service organisation that respects its customers' time.
As one industry analyst quoted in Zoom's 2026 Customer Experience Trends report put it: in the AI era, success should be measured by the value and quality of human-machine collaboration — not by speed alone, but by outcome certainty. Did the customer get what they needed? That is the only question that matters.
A Word on Trust
There is one more dimension to this that is easy to overlook: trust. The KPMG report found that integrity has become the biggest single driver of Net Promoter Score — above price, above product quality, above speed. Customers are increasingly sensitive to whether AI interactions feel honest, transparent and genuinely helpful — or manipulative and evasive.
The World Economic Forum makes a compelling point here: brands that win in the AI era will not be those with the most sophisticated interfaces. They will be those whose values travel well — through the AI layer, through the agentic systems, in every interaction where no human is watching.
This means that the design of your AI customer experience is also a brand design question. If your AI is evasive about charges, unclear about what it can and cannot do, or pushes products customers don't need — customers will notice. And they will not forgive it the way they might forgive a clunky website. The bar is higher now, because the capability is greater.
The Strategic Imperative
Let me close with the headline message, because I think it is worth stating plainly:
This is not about technology for technology's sake. It is about fundamentally respecting your customers — their time, their intent, and their intelligence. The organisations that make this shift will not just improve their customer satisfaction scores. They will build the kind of relationships that drive loyalty, reduce churn, and create genuine competitive advantage.
The organisations that don't make this shift will find themselves measuring how many pages people click through on a website that fewer and fewer people are visiting — because their customers have already moved to providers who simply answer the question.
The question is not whether AI will reshape digital customer experience. It already is. The question is whether your organisation is shaping that shift — or being shaped by it.