What does Adaptive Edge do?

Adaptive Edge is a UK strategy and innovation consultancy. We help organisations decide what to change, then make the change real: strategy built with the people who must carry it, AI adoption inside real work, and behaviour change that sticks.

How is Adaptive Edge different from a traditional strategy consultancy?

We do not hand over a deck and leave. Strategy is made in working sessions with your own teams, so the people who must deliver it own it from day one. We prove an approach in one team before asking the whole organisation to change.

Who does Adaptive Edge work with?

Global organisations across healthcare, industrials, media, retail technology and financial services, including ITAB Group, Freudenberg, Sky and Sandvik Coromant. Engagements run across Europe and beyond, remote and on site.

How does Adaptive Edge approach AI adoption?

We put AI to work inside real workflows rather than running abstract training. Teams learn on their own live work, we measure what actually changes, and we scale only what proves useful.

What does a typical engagement look like?

Most engagements combine focused research, co-created working sessions, and hands-on tools built for your context. They range from a single workshop to a multi-phase transformation programme.

How do we start a conversation?

Use the contact form on the site. We usually begin with a short call to understand the decision you are facing, and we will tell you plainly if we are not the right fit.

Your Call Centre Already Knows

By Nathan Waterhouse · 2026-08-26

"This call may be recorded for training and quality purposes." Every company says it, every customer hears it, and in most organisations the training goes no further than quality scoring. The recordings pile up, a QA team samples a handful, and the rest sit in storage: millions of customers explaining, in their own words, what they need, what confused them, what nearly made them leave. It is the largest body of customer research most companies own, and almost nobody reads it as research.

The question that unlocks the pile has a long history. Steve Baker of LTI Associates, whose Lean-era framing of it I wrote about in AI Transformation Is Not a Technology Project, puts it as: what is the nature of demand in this system? For decades, almost nobody could answer, because the measures were all inputs and the only routes to an answer, agents codifying every call or someone listening to hundreds of recordings, were priced out of reach. The question was known to matter. Answering it was too expensive to do.

AI has removed most of the expense. A model can read every transcript, classify every reason for contact, and surface the questions customers keep asking that nobody has an answer for. Speech-analytics tools have sold versions of this for a decade, and plenty of large operations run one; what those tools mostly get asked is how to deflect contacts and shave handle times. The effort has not vanished either; it has moved. The analysis itself has collapsed from a quarter's work to days. Around it, transcripts still have to come out of the telephony stack, personal and payment details must be redacted before any model reads a word, and the classifications need checking against human-read samples before anyone bets a decision on them. The consent notice that covers "training and quality purposes" does not automatically stretch to strategic research, so in the UK and Europe a data-protection assessment belongs at the start of this work, not the end. Weeks to months, then, not an afternoon. What has changed is still decisive: the cost of reading demand has fallen by an order of magnitude, and the binding constraint has moved from money to intent. Most organisations have not noticed the shift, because they are pointing the new capability at the question they were already asking.

The question decides what you find

My design-research years at IDEO were one long expensive hunt for unmet needs: interviews, observation, weeks in the field, listening for the thing people could not quite articulate. It worked, and it still works. What strikes me now is that a large share of that signal has been sitting in call recordings all along. People phone a company at the exact moment a need is unmet. The frustration, the workaround they tried first, the thing they actually wanted: it is all in the transcript, expressed more honestly than any focus group will ever manage.

Whether you find it depends entirely on what you ask. Point the analysis at "where can we save money" and it will faithfully return the contacts you can deflect and the handle times you can shave. Point it at "what are our customers struggling to get done" and different opportunities appear.

The psychology behind this is well established. In Daniel Simons and Christopher Chabris's 1999 experiment, viewers asked to count basketball passes were shown a video in which a person in a gorilla suit walks through the middle of the scene and beats their chest; roughly half never saw it. Inattentional blindness, they called it: attention recruited by one task makes us blind to what we were not asked to look for, however large and however plainly in view. A leadership team that commissions its demand analysis to count cost savings is running the same experiment on itself at company scale, with the gorilla carrying next year's growth.

Ingka Group, IKEA's largest franchisee, held both questions at once, and I wrote about the result in AI Transformation Is Not a Technology Project. One track pointed AI at deflection: their assistant came to absorb nearly half of service contacts. The other treated the same workforce and the same customer conversations as material for growth: around 8,500 call agents were retrained as remote design consultants, staffing a channel that reached 1.3 billion euros. The public record shows the two moves running side by side rather than one being discovered inside the other; what it shows just as clearly is a company that refused to read its service operation as only a cost line, and got a business as well as a saving. Two questions, held at the same time.

The contrast case arrived almost on cue. In early 2024, Klarna announced that an AI assistant was doing the work of 700 customer service agents and handling two thirds of its conversations, with projected savings around 40 million dollars. Fifteen months later the company began rehiring human agents; the assistant stayed, the all-AI posture did not. The chief executive's own post-mortem named the cause: "As cost unfortunately seems to have been a too predominant evaluation factor when organizing this, what you end up having is lower quality." Some of that is technology maturity; a 2024-era assistant in a regulated fintech was always going to strain. The part the CEO named, though, was the question. Cost was the predominant evaluation factor, and no growth story ever emerged from that body of demand, or none that Klarna has told.

The narrow question is not a technical failing; it is an organisational one. Demand analysis usually gets commissioned by whoever is under cost pressure, scoped to a brief that says efficiency, and judged against a savings target. I have watched the equivalent happen in innovation work for years. A team is briefed to find growth in one narrow lane, comes back having discovered something bigger in the next lane over, and is told: we asked about efficiency, why are you telling us about new markets? The brief forecloses the discovery before the analysis begins. Nothing about AI fixes that. It just makes the foreclosure faster.

The stance that avoids the trap has a name in the research: organisational ambidexterity, the capacity to exploit what you have and explore what you might become at the same time, rather than choosing one per planning cycle. The evidence behind it is worth stating precisely. When Charles O'Reilly and Michael Tushman studied 35 attempts at breakthrough innovation, the winners were organisations that deliberately protected the exploring, giving it its own team and senior sponsorship rather than asking the core business to do it on the side; more than 90 per cent of those succeeded, against a quarter at best for the rest. Translated to demand data, the lesson is not that one meeting can hold both questions in its spare time. The growth question needs an owner and protected time of its own, or the efficiency machine absorbs it. Design thinking trained this at workshop scale, whether we called it that or not. Diverge before you converge; hold the sponsor's efficiency brief in one hand and stay open, with the other, to the signal that points somewhere the brief never imagined. What has changed is the clock speed, the same shift I described in Why Design Thinking Needs an Upgrade for the AI Era. Ambidexterity used to be a stance you could adopt for a discovery phase and relax afterwards. When the demand data refreshes continuously, holding both questions has to be a standing habit, and the need for it is greater than it ever was in a workshop room, because the cost of reading the data one-eyed is now compounding monthly.

Sensing, not surveying

Dave Snowden, whose work on complexity has shaped how a generation of practitioners think about organisations, argues that in complex systems you cannot analyse your way to the answer in advance; you probe, sense, and respond. The sensing part has always been the hard part. His practice reached for whatever weak signals it could find, front-line conversations, staff surveys, complaints, precisely because the formal metrics arrive too late and too aggregated to sense anything with.

A call centre is a sense organ an organisation already owns. It is connected to customers at their most honest moments, it operates every day, and its output is now machine-readable at full scale. Treating it as a cost line is like owning an eye and billing it as a tear-production facility.

I have started running the same discipline on the conversations my own work produces. After a recent client programme, we put the workshop transcripts through this kind of analysis, asking not for actions and summaries but for how the room actually worked: who shaped the language the group adopted, where candour rose and fell, whose ideas carried influence. It surfaced a pattern nobody in the room had caught: a senior contributor who undercut her own ideas each time she voiced them, even as her content steered the group. Small scale, same lesson. The conversations an organisation already records will answer almost any question put to them. The deciding act is choosing a better question.

The practical version of taking it seriously is a rhythm, not a project. A monthly reading of demand: what people contacted us about, what is growing, what is shrinking, what the tone is doing, what customers asked for that we do not offer. One page, discussed in a meeting where someone has the authority to act on it. The analysis is the cheap half. The hard half is the diary time and the standing question, because a signal nobody has space to interpret is indistinguishable from no signal at all.

If your organisation records its calls, the cheapest customer research programme you will ever run is already sitting in storage, waiting for a better question. Helping teams ask it, and build the habit of acting on the answer, is the kind of work we do at Adaptive Edge.