Why digital projects fail at the implementation stage

How often digital projects fail

Anyone looking for figures on the failure of digital projects comes across the neat claim that seventy per cent go wrong. There is no evidence for that. What can be evidenced are surveys with a stated methodology, and they are sobering enough. In 2018, McKinsey surveyed 1,793 managers worldwide. Only 16 per cent reported a digital transformation that had improved performance and sustained the changes over time. In 2020, the Boston Consulting Group surveyed 825 executives and analysed 70 companies, arriving at 30 per cent successful undertakings. For the German-speaking world there is the Change Fitness Study by the consultancy Mutaree, academically supervised by a professor at the Bundeswehr University Munich. Its most recent figure is 16 per cent successful change projects, with figures between 20 and 25 per cent in the years before.

All three studies measure self-assessments, not balance sheets. Even so, they agree on one point: the causes lie with leadership, organisation and employee involvement, hardly ever with the software itself. BCG puts it this way: technology matters, but what usually tips the scales is the human dimension, meaning organisation, operating model, processes and culture.

What companies name as obstacles

At the beginning of 2026, Bitkom surveyed 604 companies with 20 or more employees. 51 per cent have difficulty coping with digitalisation. Asked about the obstacles within their own organisation, 66 per cent named a lack of time, 48 per cent a lack of money, 40 per cent lengthy decision-making processes and 37 per cent a lack of willingness to take risks. Only 21 per cent see a low willingness among staff to embrace digitalisation as an obstacle, and 13 per cent uncertainty about the economic benefits.

The order is revealing. The resistance from employees that is so often assumed comes well down the list. At the top are things that management itself controls: freeing up time, making decisions, bearing risks. Bitkom President Ralf Wintergerst concludes from this that digitalisation must be driven forward by management, and that at the same time employees must be brought along.

Which factors really make a difference

In 2015, the BPM Lab at Koblenz University of Applied Sciences, led by Professor Ayelt Komus, asked more than 450 companies to describe one successful and one less successful project each and analysed the influencing factors statistically. Asked for their opinion, the practitioners considered teamwork (83 per cent), project management (65 per cent) and team motivation (64 per cent) to be decisive. In the data, other factors had a stronger effect: conflict management, project organisation and the management of organisational change. A poor mood within the team did less damage than expected, because good organisation absorbs it. A follow-up study from 2017 with around 400 project participants adds: only 21 per cent start projects after full preparation, and 43 per cent consider the strategic priorities to be rather unclear.

McKinsey found similar patterns. Where leadership explained where things were heading and why, success was more than three times as likely. Where employees were allowed to suggest themselves what should be digitalised, the likelihood rose by a factor of 1.4, and with newly defined roles by a factor of 1.5. The consultancy Prosci analysed data from more than 2,600 change management practitioners: where change was supported excellently, 88 per cent of projects met their objectives; where it was supported poorly, 13 per cent.

Who in the organisation knows about the project

There is a banal reason why implementation so often goes wrong: many employees find out late what is coming their way. For the German Social Collaboration Study 2025, TU Darmstadt and the consultancy Campana & Schott surveyed 513 employees from Germany, Austria and Switzerland. Almost half of the employees without a management role know nothing about planned extensions of digitalisation. At middle management level the figure is 41 per cent. According to this study, tools using generative AI are mostly rolled out without any accompanying implementation measures. The authors see a high risk to project success in this lack of knowledge and low acceptance.

The Fraunhofer IAO arrived at the same picture in a 2020 case study with the Bertelsmann Stiftung. Twenty executives from 15 companies were interviewed. Technical innovations, they found, are only the beginning. What is needed is cultural change at all levels, involving as many employees as possible, a top management that models change and a middle management that carries it.

Lidl and the Elwis project

Retail shows how expensive a project can become when it does not fit the business. From 2011, Lidl worked with SAP for seven years on a new merchandise management system called Elwis. In July 2018, the group stopped the undertaking. Citing industry observers, the Lebensmittel Zeitung estimated the costs at around 500 million euros; Lidl itself gave no figure. The official explanation: the strategic goals originally defined could not be achieved with a reasonable amount of effort. By that point the system was already running in Austria, Northern Ireland and the USA.

According to reports by heise online, one core problem lay in a business accounting habit: Lidl values its inventories at selling prices, while the standard software works with purchase prices. Adapting it consumed a great deal of effort. In the end Lidl decided to continue developing its old in-house system and spoke of a decision in favour of its own system. The software ran in three countries. It simply did not fit the way the group does its accounting at a reasonable cost.

What this means for companies

For managing directors and HR managers, this shifts the question. Whether software is technically up to the job is something the provider can assess better than the customer. Whether it is accepted in the business can only be answered by the company itself. The studies provide the same touchstones for this: time set aside rather than a project run alongside day-to-day business, clear and prompt decisions, and employees who know early on what is coming and why. That includes a conscious decision as to whether the software follows the processes or the processes follow the software.

Three questions before every project sum this up: who will work with it, what will become easier, and will it be used willingly. The first question forces you to understand processes and people before a single line of code is written. The second weighs benefit and effort honestly. The third makes usability the benchmark. Software that nobody enjoys using is left untouched and ends up in the statistics as a failed project, even though it runs flawlessly.