The figures show that in terms of digital technologies, Italy is not exactly one of the most reactive countries. And so, while there is talk of Industry 4.0 and the digital revolution, we Italians have to lick the wounds of the umpteenth not very encouraging results of a DESI index which remind us – and not even too discreetly – that we are just above Romania, Bulgaria and Spain, ranking twenty-fifth in Europe. Out of 28 countries.
Among the consequences to be considered when developing Digital Transformation strategies, this generates a real cognitive bias that must be taken into account, especially when it comes to what we consider “emerging technologies”. And among the most emerging of emerging technologies, Big Data occupy a place of honor today. There is no project that does not mention them, there is no strategy that does not consider them, there is no sector that does not take them into account.
Too bad that, on leaving the interpretative perspective of someone looking at the world from fourth last position, we realize that Big Data are no longer exactly “emerging”, having already “emerged” for some time now. For so long, in fact, that emerging technologies have no longer been on the famous Gartner curve for a number of years.
“What’s happening,” in the words of Betsy Burton of Gartner (one of the report’s authors), “is that big data have moved rapidly beyond the peak of Inflated Expectations and entered fully into our lives, moreover conditioning many hype cycles for other technologies”. In short: far from emerging technology!
So, if the possible prospects of technology now travel on a global scale, we must always remember that their actual implementation depends on the definition of the global perspective on culture and the local context. A sort of “Think local, Act global” in digital guise. In other words: if a project is implemented on the basis of Big Data, problems will almost never depend on these but on how organizations are ready to understand and implement them. This is a fact that every serious designer knows well (or at least should know well).
And this, of course, generates a considerable secondary problem: close attention has to be paid to so-called best practices, a real torment and joy for marketing. A joy for marketing, because they show that something is possible; a torment for everyone else, because they induce the false belief that, in addition to that something being possible, it is easy to do. Often too easy. And they make us forget that even if something is possible in one context that does not mean that it is just as easily so in another. Another context which, by being able to rely on a technological dimension that (in theory) is the same all over the world, nevertheless experiences organizational, cultural and market dynamics that are very different and depend strongly on the local context.
And so, wanting to look at the glass half full, you realize that rather than looking at the successes of others – which depend on contextual situations that often cannot be replicated – it is often worth looking at other people’s mistakes. A sort of analysis of best practices the other way round in fact, because, after all, if everyone does well in their own way, generally speaking we all tend to make mistakes in the same way. In the final analysis, if best practices show us that something is possible, to understand how to do it we would do well to look at the “epic fails”. And given that Big Data are no longer an emerging technology, not even according to Gartner, the advantage is that many mistakes not to make are known. Mistakes from which those who want to implement a Big Data project have much to learn.
What are the most common mistakes? The following is a list in no particular order:
- Starting from technology. Given the circumstances, it could not fail to be the first point of the list. Being enchanted by technology, perhaps the latest or most advanced available, is never a good place to start. Forgetting that a Big Data project is primarily about organization, market and processes – that is, the elements that data produce – is always the best way to fail.
- Not entering the “black box”. We sometimes forget that the point is not represented so much by the availability of Big Data, that is those fast, various and quantitatively important data that the network makes available, but by the way in which these data are analyzed. So it is not enough to say, “I’m using big data analysis techniques” to put your mind at ease. We must have the courage to break the “black box” that lies behind this term and understand which methods of analysis are applied and how. In short: not all Big Data are the same. And choosing the right methodology is not exactly secondary to success of a project.
- Missing goals. Too often a technology is implemented because it is fashionable rather than because it serves to solve a real, concrete problem. But when you start from the solution rather than the problem you may not focus well on the latter and, as a result, not take the proper time to define goals. In this way, instead of the SMART model – which recalls how goals should be specific, measurable, attainable (achievable), relevant and defined in time – you find yourself with solutions in search of problems that do not focus on the goals because, by starting with solutions, you forget to define the problem. And so you fail to hit the goal (which often just does not exist).
- Lack of a “data strategy”. Wars are not won without a good strategy. Much less managing the data of a complex organization. Developing a project based on Big Data Analysis models is not simple. It is challenging and structured. Rarely do you find yourself with successful projects if they do not start with an overall analysis of the strategy used by the company to manage its data (and of those which it needs but does not manage directly: think of information from social media). Looking for specific answers rather than comprehensive solutions sometimes involves finding neither one nor the other. Lack of an overview of data management strategy is fatal for any project which contemplates the use of Big Data analysis instruments.
- Not “breaking the boxes”. Information silos in companies are not only the result of a lack of integration in information systems. They are – increasingly often – the result of the will of the different “bodies” that make up the company to maintain dominance of their information, sharing it as little as possible with the rest of the organization. So much for sharing, today management still too often starts from the assumption that knowledge is power, and that if something is exclusive you have an advantage. Personal advantage, but certainly not of the company. In short: given that inferential analysis is based on inferences among data, if you do not “break the boxes” that prevent data from being interconnected you cannot expect your project to be successful.
From lack of a strategy to the impossibility of really integrating the data available to you.. From the fallacious definition of goals to the tendency to rely on technology before models. But there is more: the history of failures related to Big Data has taught us that too often there is an underestimation of the internal marketing of the project, which, envisaging a strong cultural change, must be promoted decisively in the organization. It has shown us how training in the culture of the datum is a key element for success. It has highlighted how important it is to consider the data necessary for one’s organization not as static objects but as a true “ecosystem” which, in dynamic equilibrium, must be carefully managed so as not to alter development.
But, above all, it has taught us how implementing a successful project that has to do with Big Data means having the courage to experiment not only a new technology or analytical models, but a new way of thinking about the organization, its confines, its dynamics and its nature. And this, after all, is the most difficult choice at the base of Digital Transformation.