AI knows what your customers have said. It has no idea what they haven’t
Writing: Tom Pullen
Here’s a question that makes most leaders uncomfortable. Look back at your calendar for last week. How much of your time was spent either getting to know your customers better or making changes to your business based on their input? Not reading a report about them. Not having a meeting about them. Actually with them, or acting directly on what you learned from them.
For leaders and managers in many organizations, the honest answer is “very little.” That gap – between what companies say about being customer-centric and what their calendars say – is one of the most persistent and costly problems in business today. Managing internal stakeholders has often become the primary focus, meaning customers and their needs fall into second place; not by design, but by default. Closing the gap should be a priority – and that means reconnecting with customers directly at a human level.
What AI can’t tell you
As important as AI is for powering innovation and reshaping how we do business, there is something about it that does not get said clearly enough in discussions about customer understanding.
AI and large language models are, by their nature, backward-looking tools. They are trained on data that already exists: things customers have said, searched for, clicked on, written, reviewed, or recorded in some form. AI is extraordinarily efficient at processing that material – quickly, consistently and at a scale no human team could match. Its capability extends even to niche, less digitally connected customer groups. I recently used a generalist AI tool to surface key pain points that farmers experience with tractor tires, and it delivered remarkably accurate results, mapping closely to far more extensive – and expensive – customer research.
But here is what AI cannot do: it cannot surface the insight that has never been expressed. It cannot find the frustration that customers never thought to articulate because they assumed it was simply how things are. It cannot tell you what is hiding in plain sight – visible only to someone who goes to look.
This is not a minor limitation. While there are significant growth opportunities in innovation based on known pain points, there is an infinite stream of opportunities that simply cannot be found in existing data – ones that can lead to breakthrough innovation, strong customer loyalty and sustainable competitive advantage. These are topics that nobody has thought to ask about yet. They are sitting in kitchens, warehouses and living rooms, waiting for someone to show up.
There is a second, more structural problem with AI. When every competitor has instant access to the same tools and the same data, the resulting insight ceases to be an advantage. It becomes the lowest common denominator. The organizations that win will be those that go beyond – combining AI’s analytical capability with something it cannot replicate, and developing the innovation skills to turn what they find into tangible business growth.
AI is an incredibly powerful tool. Use it as your only tool, however, and you risk producing a polished picture of what is already known, while the most valuable insights remain invisible.
The insight that changed everything
Around 15 years ago, working in the beverage industry, I visited a consumer called Marie as part of a consumer connection exercise – where you go to meet people in their real environment and simply observe and listen. Marie was a senior woman with plenty to say. We had been talking for a while when her voice grew hoarse. I encouraged her to take a drink.
She reached for a bottle of one of our brands – and proceeded to open it with her teeth. I remember my feeling of surprise and disbelief. As a company, we had huge amounts of consumer research. Excel sheets full of data. Years of accumulated knowledge about our market and our best-selling brands. We thought we understood our consumers intimately.
Marie explained that arthritis in her lower arm meant she didn’t have the strength to open the bottle by hand. But we shouldn’t be concerned, she added – all her friends opened our bottles the same way.
Marie had never complained to the company. Never written a review about packaging. Never searched for “how to open a stiff bottle cap.” She had simply adapted – found a hack, as consumers often do – and got on with her life. Her frustration was entirely invisible to any tool trained on what consumers had chosen to express.
That single moment was found only because someone showed up, sat down, and stayed long enough for a voice to grow hoarse. It led directly to a new product concept with a full packaging redesign that addressed real consumer problems – and became one of the brand’s most successful product innovations in years.
Marie’s pain point was not hiding in the data. It was hiding in Marie’s apartment. The only way to find it was to go there.
Why asking is not enough either
There is a related misconception worth challenging: that simply asking customers questions is sufficient. It isn’t, and for a similar reason.
Customers are often unreliable narrators of their own behavior. They rationalize, simplify and answer within the limits of what they consciously know. I have listened to a consumer describing herself as “healthy” while opening a cupboard full of chocolate. I have heard of a team presenting a beautifully designed hand-soap dispenser, only to discover it takes up far too much space on a bathroom sink for anyone to buy it.
In such cases, the problem lies in the question asked. Asking more questions is a good starting point, but it is the quality of those questions that determines the value of the insight gained. Questions that introduce bias produce skewed answers. Asking whether someone prefers new product A or B assumes they want either. Better questions are grounded in real moments: What were you trying to achieve? What happened? Why did you do it that way? The goal is not to get customers to hand you the solution – that is your job. It is to get them to supply the raw material from which you can build one.
This is where I part ways slightly with the “jobs to be done” framework, valuable as it is. “Job” implies an action, yet the key is surely to focus on the desired end result. A more helpful framing centers on outcomes and obstacles – what customers ultimately want, and what is preventing or hindering them from achieving those outcomes today. Together, these form a powerful basis for successful innovation.
The antidote: immerse and inquire
A better approach to understanding customers comes down to two disciplines that are simple in principle – and consistently underused in practice.
First, immerse. Get out of the office and meet customers in their real environment, at the moment that’s most relevant to your product or service. Not in a research facility. Not through a dashboard. In context.
Second, inquire. Ask sharp, simple, powerful questions that surface real pains, real needs and real desires – without leading, steering, or projecting what you hope to hear.
Together, these two disciplines do something AI fundamentally cannot: they access the unarticulated. The unknown unknowns. They find the Marie moments – frustrations customers have adapted to so thoroughly that they no longer think of them as frustrations at all.
Immerse and inquire should be non-negotiables for understanding customers, and businesses that make them their starting point find the same things. It is easier than expected to find willing customers; the experience is genuinely enjoyable; and the team learns something actionable. Often, ‘truths’ that have circulated inside a business for years turn out not to be true at all. AI, trained on what has already been recorded, is unlikely to challenge such orthodoxy. Only proximity to real people does that.
What getting it right looks like
None of this requires a major transformation program. It starts with a decision about how leaders spend their time – and whether that decision is made visible.
I once worked with the global CMO of a major corporation who built consumer visits into every international work trip he made – going into people’s homes, observing how they lived, understanding their context directly. Not delegated, not occasional – this was simply part of how he operated. He set a tone that spread through the entire organization.
At another company, an insights manager found an effective mechanism. He placed a grid on the wall near the CEO’s office. Each member of the executive team had a box to tick when they completed their customer visit for the month. What had been easy to deprioritize became something those leaders actively scheduled – and made a priority for their teams, too.
In both cases, customer-centricity was not driven by rhetoric. It was driven by visible, consistent action. The most common reason leaders don’t do this is not a lack of interest – it’s not knowing how to start without it becoming a cumbersome project.
Start by finding real customers to speak with. Don’t overthink it. You can organize it directly – a simple LinkedIn post can work wonders. People
are more willing to share their experiences than most organizations assume. Just ensure the customer is representative of your target audience, not someone who will simply tell you what you want to hear.
Then meet them in context. Go to where they use your product or service. Observe. Ask simple, open questions rooted in real moments. And then debrief your team – not just on what you heard, but on the ‘so whats’: the observations
that warrant further investigation or immediate action. Customer insight that stays in a notebook changes nothing.
And note that this kind of approach is not only relevant for consumer-facing businesses. I have applied it in varied B2B contexts, including electrical equipment, mining, construction and materials handling.
Different environments, same underlying truth: the closer you are to your customer’s real world, the better your decisions will be.
The opportunity ahead
With AI continuing to evolve rapidly, the organizations best placed to benefit will not simply be those with the most sophisticated tools. They will be those who understand what AI is for – and what it isn’t for. Those who use it to process and scale what is already known, while investing time and energy into uncovering what isn’t.
Look at your calendar again. Did you honestly spend enough time with your customers or working on what they’ve told you? If not, that is simply a prioritization problem. It is entirely within your power to change it.
The question is not whether your organization has access to customer insight. In the age of AI, everyone has more than ever before. The question is whether you are willing to go and find the insight that doesn’t yet exist – because that is
now the only kind your competitors cannot
simply download.
Tom Pullen is CEO of Innovinco and author of Innovator: 10 simple steps to innovate with speed, scale and confidence (Practical Inspiration Publishing)
