CX Expert Says Enterprise Websites Haven’t Caught Up with AI-shaped User Behavior
Chris Gibbins explains how AI is reshaping user behavior and why CX teams must solve real customer problems, experiment often, and keep humans in control.
Chris Gibbins
Chief Experimentation Officer at CreativeCX.
Outcomes: CX leaders should evaluate AI through measurable customer and business outcomes, rejecting hype-driven investment decisions.
Workflow: AI accelerates research synthesis, prototypes, and A/B test development, while human oversight remains essential for quality.
Behavior: Enterprise websites must adapt search and content journeys because users increasingly bring AI-shaped questions and comparisons.
Guardrails: AI can confidently produce errors, requiring stronger quality assurance and flexible prompts that reduce fabricated findings.
Strategy: A problem-first, experiment-everywhere strategy helps organizations increase delivery speed without scaling ineffective customer experiences.
Chris Gibbins has led a 25-year career in user experience and experimentation, including roles like Chief Experience Officer. He currently serves as Chief Experimentation Officer at CreativeCX. And he's also the cofounder of the Experimentation London event.
We sat down with Chris to learn how AI is changing the experience for his teams and his customers. Here's what he shared.
Make decisions about AI based on outcomes, not hype
I'm a CX and Experimentation practitioner with over 25 years of experience. And now, I'm Chief Experimentation Officer at Creative CX.
Sign up to access the full article
Unlock more content to help you turn AI into leverage, design experiences that build trust, and drive business impact.
Have an account? Log In
Creative CX is a specialist experimentation consultancy with about 45 members. We help enterprises like Sky, easyJet, and Virgin Media O2 drive measurable innovation, improve customer experience, and optimize every journey through UX and Experimentation.
Crucially, we also focus on helping our clients build this capability themselves. This differentiates us from many other agencies/consultancies. We set up UX and Experimentation programs for our clients' product teams, so we are eventually no longer needed!
I work alongside a talented team of consultants, including UX researchers, UX designers, engineers, and experimentation strategists. I also work closely with some of our biggest clients and help drive the consultancy's vision and direction. This involves ensuring we stay ahead in the customer experience space, which is especially exciting now that AI has changed the landscape!
I am passionate about helping end-customers complete tasks easily while helping businesses drive measurable improvements to their digital products. I believe there doesn't have to be a conflict between customer needs and business needs in most situations — as long as Product, UX, and Experimentation are done right!
I'm also a cofounder of Experimentation London, a regular community event we started in 2024 to build and strengthen the experimentation community in London and the UK.
I am optimistic about AI's potential to improve customer experience and business performance, but equally skeptical about the hype and hyperbole in this area. Applying an experimental mindset is more crucial than ever. We must experiment with AI and make level-headed decisions based on actual outcomes, not hype!
How AI can be used in CX and development
We have incorporated AI into many of our processes to improve efficiency and enhance capabilities across development, process automation, and UX research synthesis.
We've created a number of Claude skills that help us with the latter. For example, we have a skill that analyzes large datasets from open survey questions — something that was previously impossible. Another skill that I particularly like analyzes all ideation outputs from our very large collaborative ideation sessions.
On the development side, we've significantly increased the speed of A/B test development using the bespoke framework we developed with assistive AI. This allows us to increase the rate of innovation for clients and tackle more complex experiment builds within the same budget.
Of course, A/B test development differs from 0-1 type coding — where we can expect even more efficiency savings — because we often manipulate existing code using a post-render method. This creates many complexities, so humans are still required to be in the loop.
Speaking of 0-1, prototyping is an area that has been significantly improved by AI for us. We use Claude to design fully working, coded prototypes, enabling us to build high-fidelity prototypes in 1-2 days instead of 1-2 weeks!
We're especially excited about live prototyping methods where we can build fully working concepts on top of existing websites, which we can then test with real users. Much better than the old method of stitching Figma screens together!
Why CX leaders must focus on problems and experimentation
I recommend adopting a problem-first and experiment-everywhere strategy for all Product and Customer Experience work.
Solution-first ideas are everywhere. Teams spend so much time trying to solve non-existent problems by guessing and copying competitors. A problem-first approach, by contrast, means carrying out high-quality Opportunity Discovery Research to uncover the most important customer problems across an end-to-end journey and, critically, to properly understand these problems before jumping in to solve them.
Roadmaps are no longer full of solutions. Instead, they are full of problems to solve. And the approach also tackles the problem of confirmation bias — people misusing research to back up their existing ideas.
And our experiment-everywhere approach means we believe that every team should be experimenting on every platform. The key to a brand's success lies in the potential gains from everyone's ideas. And experimentation is the key to realizing this potential because, when done well, it allows everyone to be more adventurous, test more ideas more frequently, and often test multiple ideas at the same time — all within the safety net of an experiment.
There's a reason why medicine is developed this way. We believe the same scientific method is key to creating the best customer experiences. And now, AI makes it easy!
Why AI is not the answer to everything
Sometimes people don’t need another chatbot. Instead, they might just need the telephone number to speak to a real human!
With all that said, there's so much AI noise and FOMO at the moment. AI is not the answer to everything, and that's worth remembering.
Even with all the effective use cases for AI, we ensure that it does not make final CX decisions. We also do not use it to conduct user research — though it does assist with synthesis and aspects of note-taking. We need high-quality research and data, and human experts must conduct this research to achieve that. Bad data/insights lead to poor decisions.
If AI helps you deliver a better customer experience and if it helps you solve a real customer problem more effectively than you could otherwise, then that's wonderful. But please don't see it as the solution to every problem.
Sometimes people don't need another chatbot. Instead, they might just need the telephone number to speak to a real human!
Why CX leaders must increase QA when using AI
When using AI, we must pay close attention to the quality of the final output. Because AI is often confidently incorrect.
As a result, some of the efficiency gains are offset by the extra time required for QA — whether that's in CX tasks, development, or synthesizing large-scale survey data.
Here's a tip — something that I've learned from experience. Be careful not to give overly definite instructions to AI, like "list the top 10 themes." AI will often hallucinate to reach that number!
How user behavior is changing due to AI
Enterprise websites and apps have not caught up with the public's changing expectations caused by AI use.
We recently spotted an interesting trend across several research projects, where users interacted with enterprise websites and apps very differently than they did before AI. Here are two examples:
- Internal site search: Many users type exactly what they'd type into ChatGPT into internal site searches — including product type, attributes, and other contextual aspects of their needs. This completely throws off the traditional site search and delivers terrible results.
- Working from LLMs: Many users say they now grab product URLs and move to ChatGPT/Claude, where they ask questions — for example, asking for in-depth comparisons.
Because website and app experiences haven't updated to support these new behaviors, the overall journey suffers.
Three factors that can make AI's acceleration valuable to CX leaders
…three conditions: Focus 100% on solving real problems and opportunities for your customers. Fully utilize the power of experimentation to deliver this ramped-up velocity of customer experience features, improvements, and innovations. Avoid the mistake made with personalization when it comes to AI features.
AI has already greatly accelerated team execution (design and build), and this will continue. If, like me, you care deeply about helping your customers to more easily buy, book, sign up, and get their jobs done, then this can be a very good thing — on three conditions. You have to:
- Focus 100% on solving real problems and opportunities for your customers. This means investing execution time savings into upstream research and strategy.
- Fully utilize the power of experimentation to deliver this ramped-up velocity of customer experience features, improvements, and innovations. Validate at scale; do not guess. Otherwise, you will create terrible experiences at twice the pace!
- Avoid the mistake made with personalization when it comes to AI features. Just because someone says it's AI and suggests adding it to the website, it certainly doesn't mean it will be useful. If it doesn't solve a real customer problem, it will never work, and it will likely make things worse. Roll it out as an experiment. But do not spend tons of resources solving non-existent customer problems.
Follow along
You can follow Chris Gibbins on LinkedIn.
More expert interviews to come on The CX Lead!
