New global research finds organizations keep funding AI and transformation work they already know is failing
Emergn’s study of 700 senior leaders finds the average enterprise loses 2.4% of annual revenue to transformation and AI work it could have stopped sooner, and didn’t.
BOSTON and LONDON – 1 July 2026 – Large organizations routinely keep funding transformation and AI projects that are failing, often long after the warning signs are clear, according to new research from technology and management consultancy Emergn. Politics, sunk costs, and a reluctance to admit failure keep failing projects alive when leaders know they should be stopped. The waste, the research suggests, is not inevitable.
The sums involved are large. Senior leaders estimate their organizations lose an average of 2.4% of annual revenue to transformation and AI work that fails to deliver. At a billion-dollar business, that is roughly $24M written off every year, repeated quietly, year after year. More than six in ten leaders (62%) believe their organization wastes upwards of 1% of revenue this way.
The study of 700 senior leaders in organizations of 1,000 or more employees shows the problem is not a shortage of ambition but a failure of governance. Organizations find it hard to spot failing programs early, harder still to stop them, and harder again to hold an honest view of how the work is really going.
Only 30% of leaders say that stopping an underperforming program is a normal part of how their organization works. More than four in ten (41%) act only after significant time and money have already gone in. Almost a quarter (24%) have watched a program carry on for no reason other than the amount already spent on it.
A quarter (25%) admit that decisions to stop a project are driven more by politics than by evidence. The pattern starts at the front end: on average, just 16% of new ideas are formally rejected before real money is committed.
Most leaders cannot even see the full picture. Only about three in ten (29%) could give their board a complete, real-time view of every transformation and AI program on demand; the remaining 71% would need days, weeks, or longer to assemble it. Organizations run an average of 6.6 initiatives at once, nearly half (48%) are running seven or more, and only 16% deliberately limit how many they take on. On average, around one in ten programs (10.5%) runs with no formal tracking or governance at all.
The blind spot is sharpest in the boardroom, and it is here that the two markets diverge most. Just 20% of US leaders say every program they run is formally tracked and reported to the board, but in the UK that figure falls to only 7%. It is the widest US-to-UK gap in the research, and a level of oversight few leaders would admit to in public.
The numbers also expose a gap between what organizations know and what their leaders are told. Almost a quarter of leaders (23%) say status reports paint a rosier picture than the facts support. One in five (20%) say bad news is softened on its way up, around a fifth (21%) say staff stay quiet about programs they believe are failing, and a similar share (21%) say evidence of failure gets ignored or waved away. Almost a quarter (23%) say senior leaders are reluctant to admit an AI project has failed. On every one of these measures, UK leaders flag the problem more often than their US counterparts.
None of this is a technology problem, and none of it is solved by spending more. The organizations that get real value from AI share one habit. They have introduced a product-centric operating model that gives them the discipline to focus on the most valuable ideas and, from there, allows them to maintain a live view of every initiative. They are honest about which ones are working, and they are willing to stop the ones that are not, so they can put the money behind the ones that are.
“Starting things is easy. Knowing when to stop them is the hard part, and it is where most of the money goes. Plenty of organizations can launch fifty initiatives. Very few can tell you, on any given Monday, which ones are paying off and which ones are quietly burning cash. This is a decision-making problem, not an innovation one. Companies keep funding work they already know is failing, because of politics, because of what they have already spent, or because no one wants to be the person who admits it did not work. The winners in the AI era will not be the biggest spenders. They will be the ones with the discipline to act on the evidence, stop what is not working, and back what is. That discipline is fast becoming the real competitive advantage.
Alex Adamopoulos, Chairman and CEO of Emergn
About Emergn
Emergn is a technology and management consultancy built on the belief that lasting competitive advantage comes from the way an organization works. We partner with leading global organizations, including Fortune 500, FTSE 100 and Global 2000 companies, to design and embed a work operating system that shapes how work is structured, delivered and improved over time. It brings together three dimensions of performance: process, people and technology, so that execution stays consistent and adaptable as conditions change. Our Value, Flow, Quality (VFQ) philosophy sits underneath this, helping organizations move from project-centric models to product-led, outcome-driven ways of working.
VFQ also powers Praxis, a product development platform built around how product teams actually work and learn. Through Stella, our AI product coach, Praxis gives product leaders and their teams the coaching, tools and knowledge to build products that succeed in the market, and to keep an honest, real-time view of where their investment is paying off.
Emergn has its US headquarters in Boston and its EMEA headquarters in London.
For more information, visit Emergn’s website and follow Emergn on LinkedIn.
Notes to editors
Emergn commissioned Censuswide to survey 700 senior leaders – including CEOs, COOs, CTOs, and transformation, project, and change professionals – in organizations of 1,000 or more employees across the US and the UK, split evenly between the two markets. Every respondent had been in post for at least five years. Fieldwork ran from 18 to 27 May 2026. Censuswide is a member of the Market Research Society and the British Polling Council and a signatory of the Global Data Quality Pledge, adhering to the MRS Code of Conduct and ESOMAR principles. Figures expressed as averages (for example, revenue lost, initiatives run, and share untracked) are means across all valid responses.
The 2.4% revenue figure is the mean proportion of annual revenue that surveyed leaders estimate their organizations lose to transformation and AI initiatives that fail to deliver. The $24 million illustration applies that average to a hypothetical $1 billion business and is indicative of scale, not a measured or audited total.
UK and US media contact:
Scarlett King
[email protected]
+44 7534 252 295
Portugal media contact:
Ana Sofia Pais
[email protected]
+351 910 184 186
Latvia media contact:
Nikola Kovalevska
[email protected]
+371 25 463 233
Global media contact:
Barbara Verissimo
[email protected]