Skip to main content

Posted by Kate Morris on Aug 21, 2026 12:27 PM

By Lucy Collins

Every so often, usually following the launch of another significant AI feature, I see another prediction about the death of the website.

The argument goes something like this: in the future, organisations will simply maintain repositories of information which AI platforms will scrape, interpret and serve back to us through conversational interfaces. We won't need to browse websites because AI will do the browsing for us.

Perhaps.

I’ve learnt to avoid making too many definitive predictions about where AI is heading because things have a habit of changing rather quickly. I’m also naturally suspicious of sweeping claims that a new technology is about to make an existing behaviour disappear altogether.

Usually, what happens is messier than that.

What I do think is changing, and what organisations should be paying attention to, is the mindset in which some people arrive at a website.

AI is changing the beginning of the customer journey

My husband and I are currently hoping to escape for a rare child-free long weekend to New York.

There are lots of variables involved: when to go, where to stay, what we can reasonably fit into three or four days, how much we want to spend, the restaurants we like and which neighbourhoods we’d enjoy.

These slightly fuzzy preferences are difficult to turn into a neat Google search.

Increasingly, this is exactly the sort of research I find myself starting in ChatGPT.

Rather than searching for “best area to stay New York” and working through a collection of articles, I can explain our circumstances and have a conversation. I can say what matters to us, compare different options, challenge suggestions and gradually narrow things down.

Eventually, of course, I’m going to leave ChatGPT.

I’m going to look at hotels. I’ll check prices, photographs and reviews. I’ll probably look at maps, restaurant websites, airline websites and numerous other sources before we hand over any money.

This, to me, is the interesting bit from a user experience perspective.

The website hasn’t disappeared from my journey. But by the time I arrive there, I may already know quite a lot.

Data from Similarweb suggests we shouldn’t assume this means people will always land deep within websites either. Following a change to the way ChatGPT displayed links in May 2026, around 60% of the referral traffic it generated went to brand homepages.

What has changed isn’t necessarily the page I’m arriving on. It’s the context I’m bringing with me when I get there.

I might land on exactly the same hotel homepage I would have found through Google five years ago, but I’m potentially arriving with a shortlist already formed, an idea of the price I expect to pay, knowledge of the neighbourhood and some fairly specific questions I want answered.

That creates an interesting UX challenge.

AI is another touchpoint, not the whole journey

This is where some of the conversation around AI and websites becomes unnecessarily binary.

We talk about people either “using Google” or “using ChatGPT”, when real behaviour is unlikely to divide itself quite so neatly.

Research published in 2026 into travel behaviour supports this. Researchers studying 31 million users of the Chinese travel platform Ctrip found that people using its AI assistant tended to use it during the broad research phase of their journey, particularly for exploratory questions that were difficult to express as traditional keyword searches.

Where people used both AI chat and traditional search, the most common behaviour was to move backwards and forwards between them.

That feels much closer to how people actually behave.

AI becomes another touchpoint in an already complicated journey. We might move between an LLM, Google, Reddit, social media, review sites, websites, maps and conversations with actual human beings before making a decision.

Sometimes AI might be the first touchpoint. Sometimes it might appear halfway through. Sometimes it won’t feature at all.

That last point is important.

For all the extraordinary growth in AI, we shouldn’t redesign digital experiences on the assumption that everybody is now starting their journey in an LLM.

They aren’t.

The UX challenge is to build a website that works for people arriving with very different levels of knowledge and from increasingly varied starting points.

Five things organisations should consider

1. Stop assuming you know where the journey started

Someone arriving on your homepage may know almost nothing about you.

Someone else arriving on exactly the same page may have spent the previous 20 minutes comparing you with three competitors through an AI assistant.

Both need to be able to orient themselves and move forwards without the experience making assumptions about what they already know.

2. Answer the questions people actually use to make decisions

Conversational interfaces are particularly good at allowing us to ask complicated, contextual questions.

That means somebody may reach your website looking for a specific piece of reassurance rather than a broad explanation of what you do.

Practical information becomes incredibly important: prices, limitations, timings, locations, accessibility, availability, cancellation terms and all those details marketing copy has historically had a habit of hiding several clicks away.

If somebody arrives ready to make a decision, make the information they need easy to find.

3. Recognise the website’s role in verification and trust

AI might help me decide that a particular New York hotel sounds perfect.

I’m still going to visit its website before spending any money.

At that point, I’m potentially verifying what I’ve already been told.

Does it really look like that? Is that actually the price? Is breakfast included? Can I cancel? Is the location as convenient as I’ve been led to believe?

There is some early evidence that AI-generated answers can reduce visits to source websites. One 2026 study of Google AI Overviews and Wikipedia estimated a reduction of around 15%.

Fewer visits don’t necessarily make the experience of those who do click through less important. Those visitors may arrive with considerable intent.

Your website therefore has an important job to do in creating confidence and helping them take their next step.

4. Don’t make informed users start again

We’ve always known that real user journeys aren’t neat funnels, and AI makes that particularly obvious.

If someone arrives knowing what they want, they shouldn’t have to work through the entire discovery journey designed for somebody who doesn’t.

Good navigation, comparison tools, search and clear calls to action all matter here.

More fundamentally, people need the freedom to choose their own route through the experience.

5. Don’t forget the people who aren’t using AI

Perhaps the biggest danger is replacing one assumption with another.

For years, organisations designed around the idea that journeys began with Google. We shouldn’t now decide that everybody is beginning with ChatGPT instead.

People will continue to arrive through search engines, recommendations, social media, advertising, links, review sites and direct visits.

Some people will use AI extensively. Others won’t use it at all. Even those of us who do use it won’t necessarily use it for every decision.

The goal isn’t to optimise your experience around one route into your website.

It’s to recognise that there are more routes than there used to be.

So, is AI killing the website?

I don’t think so.

It is perhaps a less dramatic conclusion than some of the predictions being made about AI, but I think it’s a more useful one.

The website isn’t dead. It isn’t becoming irrelevant.

It may, however, increasingly form part of a journey that started somewhere else, with a user who arrives knowing more, expecting more and wanting to get to their answer more quickly.

For digital teams, that raises some important questions.

What do your users already know when they reach you? What questions are they trying to answer? Which other platforms have influenced their expectations? And does your website help them continue their journey, or force them to start again?

We shouldn’t assume we know the answers.

At Web Usability, our view on AI comes back to the same principle that has always guided good user experience: understand what real people are actually doing.

AI may change customer journeys. It may introduce entirely new behaviours. It may change how people discover, compare and evaluate organisations.

Our job is to research those behaviours as they emerge, understand what they mean for real users and use that evidence to create better digital experiences.

So, with that in mind, it’s probably time to crack on with some research.

Subscribe to our newsletter

Sign up to receive regular updates straight to your inbox.