Data-driven management is gaining more and more influence in Life Sciences. Decision-making should be based on facts and no longer clouded by personal preference or opinion. However, seemingly objective facts rarely show any clear constraint that allows for only one course of action. Decisions will still be the outcome of limited knowledge, micropolitics, and compromises.
People pursue happiness, companies pursue rationality: the goal is for decisions and actions to be as logical and sound as possible. Of course, there are also such things as good feeling and intuition, but they are merely a “necessary, time-saving evil”. They do not actually meet the formal demands of management. You cannot write in the minutes of a board meeting: “The basis for our decision to increase the R&D budget is our optimistic mood.”
The idea of being able to run a business entirely based on objective facts is an age-old one. As early as 1911, Frederick Taylor published his groundbreaking magnum opus: “The Principles of Scientific Management”. These ideas seem particularly compelling in Life Sciences, an industry that was already, at least partly data-driven when no one had ever heard about concepts like big data, data lakes, data warehouses, etc. The products that pharmaceutical companies sell are based on rigorously applied scientific methods and the precise, transparent collection and evaluation of data, such as those from clinical trials. The idea is tempting to transfer this idea from the clean and aseptic universe of the R&D labs to the messy real-life world of management. The literature on data and business analytics in life sciences is almost impossible to survey. Behind all these books, white papers, and articles, behind all these shiny PowerPoint presentations and fine words lies a simplistic hope. Once a problem is clearly described, once all the facts are on the table, there will be no longer any need to get bogged down in endless discussions. Instead, you will be able to make a clear fact-based decision that is understandable and comprehensible to all parties involved.
The elusive promise of data-driven management
But is that even a realistic goal? As useful as data analytics might be in many cases, is there such a thing as objective data? Perhaps by now we have raved enough about the seemingly limitless possibilities of data-driven scientific management. Perhaps it is time to take a realistic perspective on how data are used in everyday organizational life, in contentious decisions, and in micropolitical power games.
One thing is for certain: the connection between data and the decision-making process is surprisingly difficult to make, no matter what methods of data collection are used, and no matter if the aim is to optimize purchasing, improve workflows, or better understand customers. It is an everyday experience that data look clear at first, but contradictory at second sight, or can be interpreted differently. Or that there may be a common perception of the data itself, but not on the question of whether the relationships between the datapoints are correlated or causal.
Data serve to legitimize strategic decisions
Let’s take a very concrete example. CAR T-cell cancer therapy is extremely complex, costly, and time-consuming because the CAR T-cells have to be manufactured individually for each patient. Pharmaceutical companies must secure slots in the production facilities for their patients well in advance. But how many patients will take advantage of the therapy? In theory, it should be easy to gather all the available information on this, make a prognosis, and take corresponding data-driven decisions. However, this is an extremely complicated task. The first step would involve determining the annual number of patients in a country who suffer from an oncological disease that could be treated with CAR T-cell therapy. But data on prevalence and incidence are often outdated and not very accurate. Just because a patient is eligible for a therapy does not mean they will be prescribed it. So, questions such as these also have to be answered: How do medical practitioners in the country in question view the new therapy options? For which subpopulations do they consider the treatment useful? How many medical centers are there in the country that can carry out the therapy? What is the maximum number of patients they could accept? Will health insurers interpret the approval text narrowly and initially reimburse the therapy only for smaller patient collectives? Or will they be more generous? Many more similar questions could be asked, but the problem should be clear by now. The data are too complex and too unreliable to derive a definite decision with the certainty of a mathematical equation.
Another example is the question of the future of sales forces. During the Covid-19 pandemic, sales reps could not visit medical practitioners in person. However, the data show that doctors have remained faithful to their previous prescribing practices. Furthermore, pharmaceutical companies have data that clearly show that practitioners are getting more and more information online. A good many managers in the pharmaceutical industry could combine these two observations to conclude that in the future, there will be no need for physical sales forces. Of course, opponents of this conclusion will also take a stand – and are using data too. Perhaps there are data that show that a physical sales force would have been needed, at least for the launch of new therapies. Perhaps they have also conducted a qualitative survey showing that many doctors reject online information gathering and continue to rely on the traditional conversation with a sales rep. You could even argue that the Covid-19 pandemic underscores the need for a physical sales force. After all, it was only because sales reps performed so convincingly prior to the pandemic that pharmaceutical companies were able to survive the long periods of contact restrictions.
The search for objective facts begins with subjective interests
Data and knowledge generation seems to take place via objective programs that are incapable of manipulation. But it’s not a matter of discovering real, observable entities or hidden secrets that merely need to be uncovered and understood. Each survey and each data collection is planned, conducted, commissioned, and evaluated – and at each step, the stakeholders’ decisions play a major role. They know for a fact that through data and data analysis real opportunities open up to influence the future of a company, a department or an individual’s area of responsibility.
Just look at the stages of data collection: initially, something triggers the desire for data. In describing the situation or problem, actors have already defined the perspective and therefore the premise of what should be investigated. The areas in which change is at all possible have also been predetermined. The next step is to decide which types of data to collect and which methods of measurement to use. There is no objectively simple answer in either case because different methods and ways of looking at things deliver different advantages and disadvantages.
The choice of method alone has far-reaching consequences
In all these decisions, actors are guided by their local rationality, i.e. their own interests, goals, or professional values. They each have their own core beliefs and assumptions about what is considered a reasonable course of action. Local rationalities are shaped over time through everyday work, the contexts in which actors operate, and the problems they are repeatedly confronted with. And local rationalities determine not only which data are collected by whom according to which methods. They also determine the power games around questions such as: When has data collection been completed, who will see the information first, who has the right to interpret it, and who can best reconcile their agenda with the data? The results hold sway even for those who do not approve of them. They need to respond – for example, by pointing out the weaknesses of the method or referring to other research findings that are better suited to their own agenda.
Data literacy primarily manifests itself in power games
So, what follows from all this? If departments or organizational entities attack each other using data in the way described and, in turn, ward off attacks using their own data, this need be no cause for alarm or complaint. Of course, it doesn’t mean that it would be pointless to back decisions with data. Surveys, market research, and many other forms of data collection and evaluation can provide productive surprises and irritations, lend greater certainty to intuitions and good feelings, and generate new and valuable insights.
But data literacy is not just about being able to commission market research or create PowerPoint pie charts. It means being able to deal critically with data sets, and operationalizing them for your own ends is a logical consequence. Even in data-driven management, there will still be arguments, counterarguments, compromises, and deals – and the pursuit of rationality will go on for a while yet.