Decisions Before The Code
(Or How To Wake Up From The Matrix)
Companies today are dazzled by access to data and the computational capacity required to process it. However, it is essential to understand where to begin in order to benefit from its business potential, in order to avoid the risk of getting trapped in an illusory reality where a Matrix governs our decisions.
Neo, the protagonist of the famous film Matrix, finds out his life is actually an interactive simulation, an enormous amount of data processed by artificial intelligence aimed at recreating a world that no longer exists. This first part of the trilogy by the Wachowski sisters can be divided into two parts. First, Neo becomes suspicious of the information he ‘is given’ and around which his existence revolves, until he finally ‘wakes up’ and is rescued from the Matrix. Then, our hero learns to use the information and algorithms present in the Matrix to make his own decisions, and develops abilities or ‘superpowers’ that confirm he is ‘the chosen one’ predicted by an ancient prophesy.
According to the chronology of the film, we have not arrived at that stage yet, but we definitely live in a time where the amount of information available and the computational processing capacity is unprecedented. Methodologies such a big data, analytics or machine learning, to name a few, allow us to collect and analyze huge amounts of information, and companies have taken due note of its potential for making robust decisions and achieving better results.
But where should we begin? How can the company successfully make use of this? Is the key to success encrypted in an algorithm? Certainly not. The code is really important, but it plays only a part in an intelligent decision-making process. Unless we first define the business goals we intend to improve, any data collection within reach might lead to a subjectivity bias and we might arrive at conclusions that are not really relevant to our business.
Cassie Kozyrkov, Chief Decision Scientist at Google, usually illustrates this bias with a simple and effective example: focusing on the available data first is like placing the soccer goal where the ball has fallen -it may count as scoring, but it is not necessarily valid. We are so excited about the macrodata that we forget what comes first: To decide what purpose it serves. Recent surveys indicate that a low percentage of executives across the world (around 25%) could state that their employees extract relevant information from the available data. We are looking at the answers that are within reach, instead of asking the right questions.
We propose the opposite approach based on Decision Intelligence, a discipline whose goal is to make data useful. How? By understanding beforehand which decisions we want to make and which metrics we need to leverage, leaving the collection and analysis of data for a later stage. This approach combines great volumes of data with the computational capacity to analyze it, but most importantly, it takes into consideration the business experience, fundamental to arriving at valuable conclusions. Therefore, for the analysis to be truly effective, we need to begin by asking the right questions as to which decisions we want to strengthen from the business perspective.
The right questions
Every data strategy needs to start by asking ourselves which business decisions we want to strengthen. Decisions are part of a company’s everyday life, but not all of them carry the same weight. Decision Intelligence focuses on those that have the best impact on business results and are related to commercial goals, the ones that make a difference. But those decisions also need to be recurrent, to generate a significant and updated data record that can serve to test our hypotheses. If a decision is made on a frequent basis, then the predictable value of the conclusions based on our analysis will be higher.
Therefore, following the Decision Intelligence approach, the right questions as to which decisions can strengthen will be those that meet two basic criteria: importance and frequency. To illustrate this, it is not the same to decide which brand of soap to buy than defining the services offered, prices or commercial channels (importance). Let us now think about those decisions regarding discounts made by our sales team -analysed individually their impact is low, but if we consider their recurrence and compute all the discounts that take place in a month, the impact increases significantly (frequency).
Once it has been determined which decisions we want to strengthen, we can formulate our business hypotheses and then finally decide which data analysis tools and models are more suitable for testing them.
Companies are eager to embrace the endless possibilities offered by technology to improve their business results, and they tend to start by collecting data without a clear understanding of their purpose, with the consequent risk of getting trapped in an illusory reality where a Matrix governs their decisions. At Tandem, we propose the opposite path, providing a Decision Intelligence-based approach that helps you identify the most relevant decisions for your business before thinking about the code.
Partner at Tandem.