Yes, the market is changing, and yes, search and discovery and just about every aspect of the CX is changing. But the one thing that has remained constant over the past 20+ years in digital design is the critical role of customer research.

So while it is very tempting when a new brief lands on your desk to rush off and start creating, the one lesson I have learned over the years is that without knowing who you are designing for, you cannot (successfully) design for them.

In the simplest of terms, customer research is any action taken to understand your users (whoever they may be)  before, during, and after you design for them. It sits at the intersection of psychology, sociology, and design, and it comes in many forms.

What are the types of research you can conduct?

#1 Generative research

This covers methods such as interviews, ethnographic studies, and diary research, and aims to help you understand the problem space and identify unarticulated needs.

For example, in the case of an eCommerce brand, it might involve exploring not only the market but also what is missing from the existing customer journey.

#2 Evaluative research

This is where you discover whether your designs actually work, including usability testing, A/B experiments, and cognitive walkthroughs. One area we focus on is a site's information architecture (IA). What seems perfectly obvious to one person may not be to others, so you need to prepare for that in advance. For example, if I'm on a site, I should automatically know where to find the information I'm looking for and shouldn't have to click multiple times just to get my answer.

#3 Continuous research

People’s habits and behaviours change, and so it is essential that you keep reviewing and checking on your designs. Research is never a one-and-done task; you must continue to conduct surveys, record sessions, and use analytics to keep you close to your users as the product evolves over time.

Although these methods vary in cost, timescale, and the type of insight they produce, they share a common purpose: replacing assumptions with evidence.

Here at All human we typically combine qualitative and quantitative methods depending on the product's stage and the question we are trying to answer. In early discovery, this may include interviews, workshops, and behavioural research to understand needs and shape opportunities. During design and delivery, we use usability testing, accessibility reviews, and experimentation to evaluate whether experiences work as intended. Post-launch, we rely on analytics, session recordings, surveys, and ongoing testing to continuously refine products as user needs and behaviours evolve.

First of all, let’s establish the difference between UX and CX.

• UX typically focuses on the quality of a person's interaction with a specific product or interface. For instance, consider the experience of purchasing an item online: is the process intuitive or clunky?
• By contrast, CX is broader, encompassing every touchpoint a person has with a brand—from seeing an ad to speaking with support, from unboxing a product to renewing a subscription.

Research lives at the heart of both UX and CX. While a UX designer seeks to understand how users navigate checkout flows, a CX strategist aims to uncover why customers churn after three months. Their methods and perspectives may differ, yet the core question remains: what is it like to be this person—or agent—who is attempting this task in this context?

How does research play a role in product development?

I see research as a two-way street. We get to hear from our target audience and see what they think of our designs, and they get to feed into those designs.

​Research can

• Identify pain points you'd never have thought to look for, and reveal the gap between what users say they want and what they actually do - which is way more common than most people realise.  
• Build empathy within teams, including engineers and product managers who may never interact directly with customers. And this is important because sometimes these teams can be at odds with one another, each group viewing it through its own perspective, which doesn’t always align with how people actually behave. Human behaviour is always the maverick, the unpredictable force that can make or break a new design.
• Create a shared language of evidence that makes design decisions easier to defend and align on. Designers can be a tricky bunch, as most are incredibly creative and artistic, and sometimes focus more on the design than on who it is being designed for.
• And finally, there's also a very practical business case for conducting research: products built on strong research spend less time in revision cycles. They launch with fewer critical usability issues. They retain users longer because they solve real problems rather than imagined ones. Research should not be viewed as a cost; it's risk mitigation at the earliest, cheapest moment in the process.

For most of my career, one of the chief bottlenecks in customer research wasn't the willingness to do it; it was the time. Recruiting participants, conducting interviews, transcribing recordings, synthesising notes across dozens of sessions. A solid research round could take weeks before a single insight made it into a design decision.

AI is changing that significantly.

However, let’s be clear about something: AI doesn't replace the need for research. It just makes it easier and saves time, which, for me, is where the real benefit lies.

Top uses of AI in research

Faster analysis

AI tools can process dozens of interview transcripts and surface recurring themes, tensions, and outlier signals in minutes — work that once took a researcher days.

Smarter participant matching

AI-assisted recruiting tools can screen and match participants against complex criteria, dramatically reducing the lead time before you speak to the right people.

Faster pattern detection

Natural language processing can flag emotional tone shifts in transcripts, help track how feedback evolves across product iterations, and surface what users actually feel vs what they say.

Speed for scale

AI can now analyse thousands of open-text survey responses with the nuance typically reserved for small-sample interview studies, bridging the qual/quant divide. Tools like conversational AI interviewers can run asynchronous research sessions at scale — asking follow-up questions, probing ambiguous answers, and generating structured summaries — in ways that would be logistically impossible with human researchers alone.

Importantly, and I want to stress this point: these advances don’t eliminate the craft of research—they amplify it. The designer's role shifts from transcription and tagging to interpretation and judgment. You must still ask the right questions and distinguish signals from noise. Human intelligence in research lies in framing, not filing.

And of course, we are now moving into new territory again, with companies like Brox using AI to build audiences and cut down on time in ways previously unheard of. As this article notes, their proposition is as ambitious as it is technical: the creation of a "parallel universe" populated by 60,000 digital twins of real, living human beings, each with their entire demographic profiles and consumer preferences, allowing enterprises to run unlimited experiments in hours rather than months.

It’s early days yet, and I do foresee some concerns with using this kind of tool, not the least of which is that there is an opportunity here for overlooking the unpredictability of humans, and how fickle we can be, but like I said, it’s early days yet, and so I’m sure this will be accounted for in time, if it hasn’t already!

When I started out, the only customer and experience we had to focus on was human and desktop. Then we had to start thinking about the mobile experience, and increasingly a more inclusive one that accounted for people with different abilities.

They were still all human though.

Now that’s no longer the case—this shift demands immediate attention.

AI agents, which are autonomous systems that act, browse, decide, and transact on behalf of people, are already interacting with products and services at scale. Someone recently told me that all his grocery shopping is now handled by an agent—he simply gives a detailed brief, and it handles everything else.

Which might be convenient for him, but for brands it’s opening another can of worms. Now digital experiences have to be designed for humans and agents.

Which means research needs to include both humans and agents. This has real implications for how we conduct research. Traditional usability testing assumes a human navigating a flow in real time, reading labels, making judgements, recovering from errors. Agents don't work that way. They parse structure, infer intent from content hierarchy, and can get stuck or go wrong in ways that are entirely invisible to the person they're acting for.

In response, we need to, at the bare minimum, start asking new research questions.

• How do users describe their goals to their AI assistant, and does our product's interface map onto those descriptions?
• Where do agents misinterpret or abandon flows, and why?
• What does a "good experience" look like when there's no human hand on the steering wheel?

Delaying this shift in thinking and process puts your company at immediate risk; prioritising agent experience in all UX and CX conversations is essential.

​For more on this, check out the following:

Visa CMO: AI agents are your new customers — here’s how to sell to them

AI Agents are changing digital product design