Where Transformation Really Breaks Down

business people standing around a table, looking troubled

Boston Consulting Group tracked 1,700 global transformation programmes. Only around a third accelerated growth in a way that created genuine value, 70% fell short. That figure has barely shifted in years, even as organisations have invested considerably more.


The usual explanation is that transformation is hard, which is true, but it does not get you very far. A more useful observation is that most programmes are designed at the wrong level.

Where things tend to go wrong

The strategy is usually sound. The business case is reasonable. The technology choices are defensible. The board is supportive and then the programme meets the organisation and begins to slow.

Some of this is legacy technology. The UK government’s own State of Digital Government Review, produced by Bain and Company with input from over 500 leaders across 120 organisations found that 28% of central government IT systems are now classified as legacy, with maintenance costs in some cases tripling or quadrupling those of modern alternatives. A global survey of over 1,400 senior technology decision-makers by NTT Data found a consistent picture across the private sector, with 80% of organisations reporting that outdated technology is holding back their progress.
The harder problem is the one that cannot be put on a roadmap with a go-live date. Legacy structures and legacy thinking slow transformation just as much as legacy technology does and they are considerably more difficult to address.

What cybersecurity tells us

The KPMG Global Tech Report 2026 found that more than half of UK organisations plan to increase their cybersecurity budgets by over 10% in the next twelve months, a higher proportion than any other country surveyed. Only 13% say their capability is fully scaled today and 45% report hitting blocks even with funded strategies in place.


Analysis from Osborne Clarke found that two thirds of large businesses that experienced a significant cyber incident went on to experience another. The investment went in and the incidents continued. That points to gaps in governance, behaviour and accountability that tooling alone will not close. Cybersecurity is a useful lens here because the failures tend to be visible. The same pattern, though quieter, runs through AI adoption, data modernisation and most other programmes that depend on people doing things differently rather than simply having better technology.

The layer that tends to get missed

How roles change. How decisions get made differently. How the organisation actually operates once the new technology is in place. This is the part that gets deferred, then compressed, then treated as a communications exercise rather than the design problem it actually is.

The organisations making the most consistent progress tend to treat transformation as an organisational challenge as much as a technology one. They bring the same rigour to the operating model as they bring to the technical architecture and they look for people who can work across both, which turns out to be a harder search than most programme plans account for.

You Hired an AI Specialist,  Why Isn’t Anything Changing?

Office worker sitting at a desk looking stressed about AI

It is a pattern that comes up more often than most organisations would care to admit. AI is identified as a strategic priority, the board signs off on headcount and a talented specialist is hired at a premium after a competitive process. Six months later, the outcomes that justified the hire have not materialised and the organisation is not entirely sure why.


The instinct is often to question the hire itself, but in most cases the person is exactly what was advertised. The difficulty tends to lie in everything around them.


A competitive market for a small pool of people

AI is now the scarcest technology skill in the UK. The 2025 Nash Squared Digital Leadership Report confirmed what most people recruiting in this space already sense. Top candidates are fielding multiple offers, rates continue to climb and the competition for proven talent is significant. This is happening alongside an overall market that has cooled considerably, with UK vacancies down 43% from their 2022 peak. There is real competition, but it is concentrated around a relatively small number of people.

Organisations work hard to win the hire. The challenge is what happens next.


The structural problem

From the inside, it tends to look like this. The mandate is not quite clear. Data access sits behind governance processes designed for a different era. The authority to make meaningful decisions rests with people who are not close enough to the work to exercise it well. The role has landed in a function that treats AI as a support capability rather than something central to how the business operates.

Research from IMD Business School found the average tenure of a Chief Digital Officer is 31 months, the shortest of any C-suite role and that the high rate of departure is rarely explained by a lack of competence. The role exists. The conditions for it to succeed often do not. A similar dynamic appears to be emerging one level down, across the AI and data functions being built across UK organisations today.

The KPMG Global Tech Report 2026 found that only 3% of UK technology executives consider their AI capability fully scaled, despite significant investment. Almost half said they were encountering blocks even where the strategy was funded and had board backing. That points less to a talent problem and more to an organisational design one.


What organisations getting this right tend to do differently

They do the structural thinking before the talent search begins. Before the job specification is written, they are asking what this person will actually have authority over, what data they can access from day one and who in the leadership team genuinely understands what good looks like in this role.

A capable person in the wrong structure will not deliver what was promised. That is not a recruitment failure, it is a design decision that was made, often without realising it, long before the candidate accepted the offer.