The Layer and the Core: Why Digital Transformation So Rarely Touches the Business Model
The customer meets the future. The enterprise stays in the present.
Executive Summary
For several years now the language of digital transformation has dominated the boardroom, and the spending has followed the language. Yet a curious pattern recurs across sector after sector: the customer’s experience is transformed while the economics of the enterprise are left almost exactly where they were found. A new mobile app, a data lake, a customer-experience platform, an innovation lab in a converted warehouse — and, underneath all of it, the same business model, the same margin structure, the same fundamental way of making money that existed before the programme began.
This essay argues that the pattern is not a failure of understanding, nor of ambition. It is the predictable output of how digital initiatives are funded, governed, staffed, and measured. Each of those four mechanisms quietly converts a question about the business model into a question about technology — and technology is what duly gets delivered. The essay traces those forces, then gives the strongest case for leaving the model alone, that layering is prudent sequencing rather than avoidance, and draws the line that separates the two. What the pattern finally reveals is less about digital than about how large organisations change at all: they will reshape almost everything they can see and touch before they will reopen the one question that made them what they are.
The app was beautiful
The app was beautiful. It had taken fourteen months, a team that grew from nine people to something over forty, and a budget that everyone had learned not to say aloud in one number. On the morning it went live the programme director stood in front of the board with a slide that showed a five-star rating, a download curve bending pleasingly upward, and a Net Promoter Score that had moved further in a quarter than the business had managed in the previous three years. There was, quite rightly, applause.
Eighteen months later the same organisation was doing the same things, for the same customers, at the same cost, and making its money in the same way it always had. The app had changed how it felt to be a customer. It had changed almost nothing about what the enterprise was. The cost-to-income ratio, the metric that actually decided whether the institution thrived or was quietly absorbed by someone larger, sat within a rounding error of where it had been before the first line of code was written.
I have watched this sequence enough times now to believe it is not an accident, and not a matter of the wrong people or the wrong vendor. The most striking feature of the digital transformation programmes of these last few years is how reliably they transform the surface and spare the core. We should be curious about that reliability. Accidents are random; this is a pattern. And patterns have structure underneath them.
The question worth sitting with is not why do some digital programmes fail to change the business model? It is the opposite: why do so many succeed at everything except changing the business model? Something is protecting the core with great consistency. That something is worth naming.
What “digital” quietly came to mean
It is worth being precise about the confusion at the heart of the term, because the confusion is doing real work.
“Digital transformation” fuses two ideas that point in very different directions. The first is digital — a set of technologies and channels: mobile, web, data platforms, analytics, the cloud, application programming interfaces. The second is transformation — a change in what an organisation fundamentally does and how it captures value. The phrase invites us to assume the first delivers the second. Buy the technology, the grammar suggests, and the transformation comes bundled in.
But there is no law of nature that makes this so. A bank can put a superb application in every customer’s pocket and remain, in every respect that matters to its economics, a branch-and-spread business with a nicer front door. A retailer can build a genuinely excellent website and remain a property company that happens to sell goods, its cost base still dominated by leases signed in a different era. In each case the digital layer is real, expensive, and even admired — and the business model beneath it is untouched. The customer meets the future. The enterprise stays in the present.
The gap between these two things is where the money goes and the value does not. And the reason the gap persists is not that leaders fail to see it. Many see it perfectly well. The reason it persists is that four separate mechanisms inside the organisation each, independently, push the programme back toward the layer and away from the core.
The first force: how the money is allocated
Follow the money and the pattern begins to explain itself.
A digital initiative, at the point it needs funding, has to become a business case. And a business case, in an established organisation, has a fixed shape: it must show a return, within a defined period, against the existing model of the business. That last clause is the quiet trap. The template itself assumes the model as a constant and asks only what the investment will add to it. You may propose to improve the model. You may not, in the standard case, propose to replace it — because there is no line on the form for the revenue you would destroy on the way to the revenue you would create.
So the projects that clear the gate are, systematically, the ones that layer onto the existing economics rather than disturb them. An app that lifts retention: fundable, because the benefit lands inside the current model. A platform that would let the firm make money in an entirely new way, at the cost of cannibalising the old way for three years first: unfundable, because the business case cannot hold a loss it is structurally forbidden to name. The capital allocation process is not neutral about the business model. It is actively, if unintentionally, biased toward preserving it.
“The template that funds transformation is the same template that protects the thing transformation is meant to change.”
This is why so much digital spend takes the form of capital projects with the texture of IT delivery — scoped, gated, tracked against a return that is, by construction, a return to the incumbent model. The programme was never asked whether the model was right. It was only ever asked to make the model perform a little better while wearing new clothes.
The second force: where digital sits in the organisation
The years just past gave us a new figure on the executive floor: the person made responsible for digital. Sometimes a chief digital officer, sometimes a director of digital, sometimes simply the leader of “the digital team.” Wherever the title landed, one feature was almost universal, and almost fatal.
The role carried responsibility for digital and no authority over the business model.
Consider what that means in practice. The person charged with dragging the organisation into the future typically owned a channel, a budget, and a team — but not the pricing, not the product economics, not the branch network or the store estate or the underwriting rules, not the profit-and-loss account of the core business. They could build. They could not re-decide. And an organisation’s business model does not live in its channels; it lives precisely in the pricing, the products, the cost base, and the P&L that the digital leader had been carefully placed just to the side of.
- The digital unit becomes a place where new capability is built and old power is not touched.
- The people with authority over the model are, by design, not accountable for the digital agenda.
- The people accountable for the digital agenda are, by design, without authority over the model.
The result is an elegant division of labour in which everyone is doing their job and no one is positioned to change the business. The structure guarantees the outcome. You cannot transform what you have organised yourself to leave alone.
The third force: what gets measured
The third mechanism is the most seductive, because it wears the costume of rigour.
Digital programmes are measured, heavily, and by numbers that feel modern: downloads, active users, engagement, session length, Net Promoter Score, star ratings, conversion. These are not worthless. But notice what they share. Almost none of them is an economic measure of the enterprise. They describe the intensity of the customer’s relationship with the surface; they are largely silent about whether the firm makes money in a better way than it did before.
A programme can move every one of those numbers in the right direction and leave the cost-to-income ratio, the return on capital, the unit economics, exactly where they were. And because the modern numbers are the ones on the programme dashboard — because they are the numbers the programme was set up to move — everyone can be honestly, energetically busy improving metrics that were never the point. The dashboard becomes a device for not noticing that the model has not moved.
I use the word honestly deliberately. There is no cynicism required for this to happen. Give a capable, well-motivated team a set of engagement targets and they will hit them. The failure is not of effort. It is that the measurement system quietly substituted a question it could answer — are customers engaging? — for the question that mattered — are we now a fundamentally better business? — and no one felt the substitution happen.
The fourth force: the core defends itself
The three forces so far are mechanical. The fourth is closer to instinct, and it is the strongest of them.
An incumbent’s business model is not merely how it makes money. It is the accumulated settlement of every internal interest: the division that owns the profitable product, the region that owns the branch estate, the function whose headcount is justified by the current way of working. To put the business model on the table is to reopen all of those settlements at once. And the parts of the organisation that do best under the current model are, not coincidentally, the parts with the most power to defend it.
So genuine business transformation triggers what can only be described as an immune response. It does not usually take the form of open refusal. It takes the form of a thousand reasonable objections, each locally sensible, that in aggregate ensure the model survives: the risk that cannot be accepted this year, the customers who mustn’t be disrupted, the regulatory conversation that isn’t ready, the systems that can’t be touched until the other programme finishes. Every objection is true. Together they are a wall.
The digital layer survives because it threatens no one. It adds a channel without subtracting a privilege. That is precisely why it is so much easier to fund, staff, measure, and complete than the transformation it was supposed to be a vehicle for.
This is the deepest of the four forces, and it explains why the other three are so rarely overridden even when leaders see straight through them. The layer is the path of least resistance not because people are timid but because the core is genuinely, structurally defended by those who benefit from it. Digital transformation without business transformation is, in this light, not a bug in the programme. It is the organisation working exactly as an organisation is built to work: protecting its own settlement.
The strongest case for leaving the model alone
It would be too easy to stop there, with the incumbent as villain and the timid programme as tragedy. Honesty requires that the other side of this be put as strongly as it deserves, because there is a real argument that the layering approach is not cowardice but wisdom.
The argument runs like this. A large, going concern has obligations that a start-up does not: to customers who depend on it, to employees, to the stability of a system it is part of. Betting the whole enterprise on an untested new model, in the name of transformation, is not courage; it is recklessness with other people’s money and livelihoods. The prudent path is to build the new capability at the edge, learn from it, let the evidence accumulate, and shift the model only when the shift is de-risked. Under this reading, the app and the data platform are not avoidance at all. They are real options — modest investments that buy the right, without the obligation, to move the core later, when the future is legible enough to bet on.
This is a serious argument and I do not think it can be dismissed. Sequencing genuinely is intelligent. Not every organisation should reinvent its economics on a hunch, and the graveyard of firms that bet the business on a premature vision is at least as crowded as the graveyard of those that layered too timidly.
Where the argument fails — and the line between the two
But the real-options defence contains a test, and most programmes fail it.
An option has value only if someone intends, under defined conditions, to exercise it. A modest investment at the edge is sequencing if there exists a person with the authority to change the model, a set of conditions under which they have committed to do so, and a genuine intention to act when those conditions are met. It is avoidance dressed as sequencing if none of those three things exists — if the “option” will, in practice, never be exercised because no one owns the exercise, no trigger has been named, and the edge investment has quietly become the whole ambition.
The distinction is not philosophical. It is diagnosable, and it can be diagnosed with three questions:
- Is there a named individual with the authority and the accountability to change the business model — not the channel, the model — as a result of what the digital programme learns?
- Have the conditions under which the model will change been written down in advance, or does “we’ll see how it goes” stand in for them?
- When the edge investment succeeds, is the plan to feed that success back into a decision about the core — or simply to build another edge investment?
Where the answers are yes, the layer really is a bridgehead, and the sequencing is sound. Where the answers are no — and in my experience they are usually no — the language of options is being used to make permanent avoidance sound like patient strategy. The tell is always the same: everyone can describe what the app will do, and no one can describe who would change the business and when.
“Sequencing has a name and a date. Avoidance has a roadmap that never arrives at the core.”
What this reveals about how organisations change
Step back from digital altogether and the pattern turns out to be about something larger.
The four forces — capital allocation that assumes the model, structure that separates authority from the digital agenda, measurement that substitutes engagement for economics, and a core that defends its own settlement — are not peculiar to technology. They are the standing machinery by which established organisations metabolise any change. Digital simply gave that machinery an unusually clear stage, because digital arrived loudly enough that everyone had to respond, and expensively enough that the response could be measured. What we learned watching it is how an organisation behaves when it is asked to change: it will do a very great deal, spend heavily, work hard, and reshape everything within reach — right up to the edge of the one question that would actually change what it is, and there it will stop.
Which means the counsel that matters is not do more digital. Almost everyone did enough digital. The counsel that matters is to notice which question your programme has been organised not to ask, and to see clearly that the organising was the point. If the business model is not on the table — if no one owns changing it, no condition would trigger changing it, and no measure would tell you whether it had changed — then however beautiful the app, the transformation was decided against before it began.
The app was beautiful. That was never in doubt. The question was always whether anyone was allowed to change the business, and the honest answer, in most of the programmes I have watched, was written into their structure long before the first review: no.
It is worth being fair, in closing, to those who ran these programmes. Many understood all of this perfectly. They layered because layering was fundable and transformation was not, because they held a channel and not a P&L, because the numbers on their dashboard were the numbers they had been given. They were not the reason the core went untouched. They were, most of them, doing excellent work inside a machine built to spare the very thing they had been asked to change. The failure was never theirs to fix alone — and recognising that is the first step to building programmes where the question of the model is, for once, allowed onto the table.