Again and again one hears that it is now finally time to dispose of digitalisation as a cross-sectoral, transversal issue. We have to develop digitisation concepts in each discipline, from the logic of the discipline. The time is over that we consider digitisation as a topic in and of itself. In other words, digitisation should be developed bottom-up in the future and not transversally as before. The latter is still practised by the big Silicon Valley corporations, but in many sectors they will be seen as the enemy that you don’t want to let into your business.

The bottom-up approach develops digitalisation from within the existing thinking logic of a sector and/or discipline. This actually has many advantages, in particular it uses the experience and expertise of the respective sector. Thus, it mostly addresses core problems and uses existing competences. On top of that, the resulting concepts and products are easy for the professionals to communicate. However, the bottom-up approach also has disadvantages. The methodological knowledge of digitalisation is ignored and the potential of digitalisation to achieve a better design faster with alternative approaches is not used. The consequence of the bottom-up approach without transversal digitisation knowledge is that many digitisation projects are tackled the wrong way round. For example, data is collected first and then a good design is created in one step, instead of the design being created from the beginning to support the collection of data and its own step-by-step development. Or the digital transformation project is organised as a single, large multi-stakeholder project where the intention is to involve everyone, rather than as a series of many smaller projects, each with so few stakeholders that they could actually be successfully involved.

What happens when knowledge is not retrieved

The transversal approach avoids such design disasters because it draws on previous project experience. It also has other advantages, for example, methodological knowledge in data science can be excellently used transversally. Unfortunately, the transversal approach also has serious disadvantages in practice. Namely, domain knowledge is often ignored and experiences are used generically across domains, but the domain-specific pre-digital experiences are usually not taken into account, even if they could be translated into digital contexts. So far, some transversal projects have therefore failed prominently because of the insufficient inclusion of domain knowledge. Moreover, the transversal approach has repeatedly shown its ugly face when it came to domain-specific bottom-up initiatives. This is because the big corporations try to prevent domain-specific innovations from emerging in the first place in areas that are on their strategic roadmap. While this is understandable from a corporate perspective, it considerably slows down the digital transformation of the economy and society. In terms of success, however, the bottom-up approach has so far had an even worse record than the transversal approach – and this despite the fact that data science, transversal methodological knowledge is increasingly being used for this purpose. The bottom-up initiatives imitate the large Silicon Valley corporations to some extent in the area of abstract methods, but rarely if ever in terms of corporate culture. The consistent to extreme factual orientation is rejected, as is the conviction that technology is decisive. However, it is precisely the technology and the extreme corporate cultures that have made the very big successes in Silicon Valley possible, in addition to many flops.

Similarities and differences with democracy

Some politically-minded readers will probably think of today’s dilemma of democracy in this portrait of the two competing views: our democracy faces the challenge of reconciling cosmopolitan and globally-minded universalists with the political tribal societies that define themselves by thinking they know the truth about the world, despite the supposed lies of those in power and the traditional media. The transversal approach has strong parallels with universalist thinking, while the bottom-up approach resembles the thinking of tribal political societies. Of course, a disciplinary way of thinking has more legitimacy than an ideological view of the world – that is beyond doubt – but the rejections of universalist thinking are actually the same in the digital as in the political. The challenges of the digital economy, however, are different from the challenges of democratic policymaking. While in politics the integration of all is sometimes even more important than meaningful action on the matter, in the economy not only different actors of a country compete against each other, but also states/communities of states against states/communities of states. The latter makes it urgent for us in Central Europe, for example, to consider whether we do not need to substantially develop our innovation ecosystem in order to create more output. The lack of a transversal exchange on digitalisation is economically disadvantageous and reinforces the division of society. Even without the new trend towards bottom-up digitisation in sectors, the ACTUAL state of our innovation ecosystem gives us little hope that we can maintain our economic leadership in the future. And this is not only due to the much-discussed EU research funding, which is not effective enough.

Finding the right method

If we look at the situation with some distance, we see what almost all discussions overlook. Something important and central is missing. The methods with which a transversal approach can use the knowledge and experience of the respective sector and the disciplinary expertise central to it. These methods not only do not exist, they can only exist to a limited extent. More precisely: There are and need to be countless methods, all of which may be necessary in the case and whose use in each case results from the substantive engagement with the digital transformation and all its inherent ambiguity. However, these methods are so numerous that one cannot speak of methods in the scientific sense.

1. This means, firstly, that in a concrete case perhaps only five aspects are decisive for success and for each of these aspects there are only two or three methods, whereby at best a new method must be developed ad hoc for one aspect. But in the sum of all concrete cases, the relevant aspects are too numerous for us to be able to count them on our fingers and toes, and existing methods must always be adapted or even newly invented. Thinking and acting bottom-up is uneconomical, among other reasons, because it requires much more reinvention and thus creativity than a transversal approach that uses existing experiences from other areas.
2. But it also means, secondly, that even the selection of the decisive aspects – there are not always five, of course – and the methods that go with them is rarely clear-cut, although a wrong choice can spoil everything and often – the list of concrete failures is long – blocks progress for many years to come. This is why managers can only lead digitisation projects with an understanding of content. This is why the popular lists of the five most important aspects are only of limited use, even if they are useful for banal projects in everyday business. Many key digitisation projects are not trivial – especially in the public sector, in large corporations and when completely new services are brought to market by start-ups.
3. But it also means, thirdly, that the methods must be applied precisely despite their multiplicity and the respective ambiguity as to which method is needed. Otherwise, one fails precisely because of the right method. The popular faking of agility in practice always provides sad illustrations of this. So does data science practice. Diversity and ambiguity force accuracy! This is exactly why some corporate boards and startups begin their meetings with silence. This is exactly why they try to stop all kinds of show. This is exactly why they take a long time to make fundamental decisions, but don’t take their time in implementing them.

Or in the holistic way

The alternative to the bottom-up approach and the transversal approach is a lean – in the true sense of the word fit – holistic approach. This approach is not solutionistic, as is popular in Silicon Valley, because the inherent complexity of the world is too high for that. But it does not solve global problems either, but keeps the complexity of the individual projects low and observes their interaction on the basis of models developed for this purpose. For the projects themselves, it uses – very briefly formulated – domain expertise, generic methods, competence in aspect vision, transversal digitisation experience and people’s joy in playing and experimenting. This is more than the sum of the two and is still more efficient. So much more efficient that it is also substantially more effective in the end result!