The Future of Insights Isn’t an Entirely New Question
What changes when a profession returns to familiar challenges with new tools?
This essay was developed through original research and analysis with AI assistance in the research, drafting and editorial process.
Sí. Eliminé los nombres de sitios que habían quedado sueltos y los integré en el texto. También vinculé los títulos en la sección final de fuentes y quité los elementos de interfaz de Medium que aparecían en el texto pegado.
Mantuve el ensayo tal como estaba. Para reconstruir los enlaces que se habían perdido al copiarlo, contrasté las referencias con las publicaciones originales, entre ellas las de BCG, BMJ y MRII. bcg.com
The Future of Insights Isn’t an Entirely New Question
What changes when a profession returns to familiar challenges with new tools?
This essay was developed through original research and analysis with AI assistance in the research, drafting and editorial process.
Image caption: More signals can help us see more. Understanding what they mean still takes work.
In 2017, a report on Customer Insights argued that teams needed to bring information together, make it easier to access, and play a more active role in business decisions.
It mentioned digital data and self-service tools, too.
It sounds like a conversation we are still having.
The report, published by BCG and based on research conducted with Cambiar and Yale, described a familiar challenge: many organisations had valuable knowledge about their customers but struggled to bring it into decisions. It also identified teams that were already working more closely with other parts of the business.
It would be easy to say that nothing has changed.
But that would mistake a recurring ambition for an unchanged practice.
In 2017, making information accessible could mean creating dashboards or repositories of previous studies. Today, it can also mean that someone asks a question in natural language and receives an initial synthesis drawn from several sources.
The ambition is familiar. The conditions for pursuing it may be changing.
That distinction matters. A new way to produce an answer does not necessarily mean an organisation has changed how it understands a problem or makes a decision. But if we focus only on the familiar ambition, we may miss what is changing: who can ask the question, who can arrive with a first interpretation, and whose perspective gets considered.
That raises a more useful question than whether Insights needs to reinvent itself: what are we still trying to do, and what is becoming different about the way we can do it?
The answer before the meeting
Imagine a brand review.
The consideration score has fallen in one market. Before the meeting begins, someone has already checked search trends. Another colleague has looked at distribution. The agency has a view on a competitor’s campaign.
By the time the team sits down, there is no shortage of possible explanations.
The harder question is whether everyone is answering the same one.
The brand tracker shows a movement. Search data reveals growing interest in a competitor. Sales data shows that the brand is available in fewer places. Each source contributes something, but none tells the whole story.
A platform that brings signals together can help a team find these pieces sooner. It might even surface a relationship no one had considered.
The value of a new source is not only that it can give us an answer faster. It may bring an unfamiliar question into view — one the tracker was never designed to ask.
A cluster of reviews might reveal a difficulty after purchase, prompting the team to look beyond awareness and consider whether the product experience also deserves attention.
Stan Sthanunathan, formerly of Unilever and Coca-Cola and now associated with i-Genie, has argued for paying closer attention to what people search for, discuss, and review online. His company offers tools that bring some of those signals together.
A survey can miss something because no one thought to ask about it. An online signal can reveal an unfamiliar question — but it does not automatically tell us how widespread that experience is or what caused a movement in a brand tracker.
Connecting sources to understand a situation did not begin with AI. Insights teams have long brought together what people say, what they do, and the conditions in which they make choices.
The question is what changes when more people — and more systems — can participate in that work.
Sometimes sources point in a compatible direction. Sometimes they illuminate different parts of a situation. They can also appear to disagree: searches for a brand may rise while its consideration score remains stable.
That difference is not necessarily an error to be removed in order to produce a coherent story. It may reveal that the sources are observing different things. Mixed-methods research has long paid attention to both agreement and disagreement between findings.
Connecting the dots can also mean recognising that we do not yet know how they fit together.
Who gets to ask?
Return to that brand review. Before the meeting, the brand manager consults a platform and receives a synthesis suggesting that a competitor’s campaign may explain the decline.
The answer could enrich the conversation. It might help the team arrive better prepared or explore a relationship no one had considered.
But the starting point is different. Insights is not always the first place the question goes. Other functions can arrive with data, hypotheses, and interpretations of their own.
That does not make the work of Insights unnecessary. It can change where and when the function contributes: helping to examine sources, putting alternative explanations on the table, or connecting an answer to a particular decision.
Nor does it mean a brand team needs a specialist for every question. If several reviews point to an error in a product description and someone confirms it, there may be enough information to correct it. The action is specific and reversible.
Reallocating a major investment based on a relationship the team does not yet understand is a different decision.
The useful question is not simply, “Do we have enough information?” It is:
Enough to decide what?
A signal may be enough to begin exploring. A finding may justify a small test. A decision with greater consequences may require a different combination of evidence.
Researcher Ray Poynter has suggested that, when considering AI in market research, the starting point should be the problem we are trying to solve rather than the method traditionally used to solve it.
That challenges an assumption that can hide inside the idea of continuity. If the question remains, what makes us certain that the existing research process — or the existing role — is the best way to address it?
A new approach should be judged against the question it claims to answer and the evidence available to support it. Preserving an established process is not an end in itself.
The fact that an organisation still needs to understand people does not tell us which team will do that work, how much of it will be shared, or what an Insights function will look like.
Get Roger Federico Rybak’s stories in your inbox
Join Medium for free to get updates from this writer.
Those are not questions the technology answers on its own.
A faster answer can change a decision. It can also arrive faster into a process where the decision has already been made.
A tool enters an organisation
When a new technology appears, it is easy to talk as though its capabilities determine its effects.
If it can bring information together, the organisation will have integrated knowledge. If it can answer questions, decisions will be faster. If it automates part of the analysis, the team will have more time to think strategically.
Each of those outcomes is possible. None is guaranteed by the tool.
Two teams can be given the same platform and end up working in different ways. One may use it to make existing reports faster. Another may use it to let more people explore questions before a decision is made.
Organisational researcher Wanda Orlikowski studied how a technology became part of everyday work, drawing attention to the adjustments and improvisations that occurred as people used it and changed their routines around it.
Her perspective invites us to look beyond the moment a platform becomes available. Who uses it? What can they question? Which decisions are still open when the answer arrives?
Bringing a platform into an organisation is one decision. Making its sources usable, its outputs understandable, and its findings available at the right moment is a longer piece of work.
For those of us working in Insights, a new tool can bring two impulses at once: curiosity about what it makes possible and a desire to understand what it changes about the value of our own work.
Neither feeling tells us what the outcome will be. To understand that, we have to look at the processes, resources, and decisions the tool encounters.
What we know — and what we have yet to ask
An organisation that can search years of research may be better equipped to answer questions it already knows how to ask.
That could address a familiar frustration. Many organisations have investigated similar questions more than once because what they learned before was difficult to find, compare, or use.
A tool might reveal that a need had already appeared in several markets or that an attractive hypothesis had been challenged before. It could help connect findings that had been kept apart.
But what happens to the questions that were never included in those studies?
Researcher James March explored a related tension: organisations need both to make use of existing knowledge and to explore new possibilities. Both require time and resources. Refining what is known can offer more immediate returns; exploration can take longer and cannot guarantee what it will find.
So it is worth asking what happens to any time that easier access makes available. Does it go towards answering more familiar questions — or exploring something we do not yet know how to ask?
The same tool could support either. It might reinforce existing categories, or it might help reveal their limits. That is something to observe in practice, not decide in advance.
An organisation’s memory matters for what it preserves and for what it helps us notice is missing.
What the past cannot tell us
The 2017 report helps us recognise that some ambitions are not new. But it cannot tell us what organisations will do with today’s tools. Nor can it tell us that the same questions will remain central ten years from now.
One trap of looking back is reconstructing events as though the path had been obvious all along. Once we know the outcome, we can give too much weight to the signals that appear to have anticipated it and forget the possibilities that never materialised.
Foresight offers a way to explore those possibilities. It does not necessarily seek to predict a single future. It can help people examine different conditions and think more clearly about decisions in the present.
For Insights, that might begin with a question: what would change if more teams could directly consult the available evidence, but the spaces where decisions are made remained organised much as they are today?
Or what if that evidence depended on a small number of platforms whose criteria were difficult to examine?
These are not predictions. They are ways to examine what we currently take for granted.
The questions do not stay unchanged
When we talk about the future of Insights, it is easy to produce a list of capabilities: learn new tools, work with more sources, develop influence, become more agile.
All may matter. None exists in a vacuum.
A team can have access to continuous signals and still arrive too late to influence a decision because the budget has already been committed. It can have strong research and no opportunity to question the original brief. It can also discover that a shared platform brings more people closer to consumers and makes it possible to investigate questions that previously went unanswered.
In a 2026 survey by the Market Research Institute International, approximately half of the 498 professionals consulted said they used AI regularly, while around one in ten said it was fully embedded in their standard workflows. The survey is not representative of the entire industry, but it is a reminder that using a tool and transforming the work around it can be different processes.
The question is not only what Insights needs to learn. It is also which decisions are still open when evidence arrives, who can challenge an interpretation, and what the organisation learns after it acts.
Back in the brand review, the team may decide it already knows enough to run a small test. Or it may realise that the decision it was about to make depends on an explanation no one has yet examined.
Either way, the first answer is not the end of the conversation. It helps determine what the team needs to do next.
Insights cannot create that space alone. If the direction is already set, even a strong answer may arrive too late to change it.
It is understandable to look for reassurance in what remains — or excitement in what is new. But neither tells us, on its own, what the work now requires.
Sometimes the words stay the same while the practices change. At other times, the tools and the language of transformation change while familiar ways of working and deciding persist.
To distinguish between them, we have to look at what happens in practice: who asks the questions, how answers are constructed, and which decisions can change because of them.
Looking at an old question does not return us to an old way of working. It can help us recognise what is different now — and what still needs to change.
Perhaps the task is to recognise the questions that remain — and notice when they are becoming different questions.
Sources
- BCG, Cambiar and Yale Center for Customer Insights, How Customer Insight Can Be a Powerful Business Partner, 2017.
- i-Genie.ai, From Asking to Listening: Rethinking Consumer Insights in an AI-Driven World.
- O’Cathain, Murphy and Nicholl, Three Techniques for Integrating Data in Mixed Methods Studies, 2010.
- Ray Poynter, How AI Can Improve Market Research Decisions.
- Wanda Orlikowski, Improvising Organizational Transformation Over Time, 1996.
- James G. March, Exploration and Exploitation in Organizational Learning, 1991.
- OECD, Using Foresight Methods to Adapt Development Co-operation for the Future, 2018.
- Market Research Institute International, 2026 State of the Market Research Industry, 2026.
