CRACKS of technology’ is a weekly series of interviews, through which we want to give voice to those IT professionals who are absolute geniuses of technology in Spain. We want to hear from them, to know and recognize the work they do in these companies; to know what they are passionate about and what advice they have for those who will come after them.

---

 

The keys to solving business problems are hidden in company data. Knowing where to look, selecting the type of statistical model that will give you the answers you need, and how to interpret those answers is the foundation of data science.

Today we talk to the head of Data&Analytics at Overlap, a consulting firm specializing in business and human resources management, about how to transform an organization into a data driven company [empresas dirigidas por datos]. Because making decisions based on data, and not [sólo] on intuition, is not as simple as it seems.

Miguel Sáez has extensive experience in analytics and business consulting, where he has worked on projects for different companies and sectors with a focus on the strategic definition to obtain results through the analysis and exploitation of information. This has brought him into contact with the business and development areas, as well as with the technology and infrastructure areas. From his experience, he tells us some keys about data analytics, one of the job fields with the greatest potential for the future.

Miguel Sáez Avilés, Data&Analytics Manager at Overlap.

Q.- What does it mean for a company to do Data Analytics?

A.- Very simple, Data Analytics is the ability to exploit and analyze the available information to help make decisions in a more objective and agile way. And not only considering intuition, perception or personal experience. It is about avoiding falling into the prejudices that – let’s face it – we have been acquiring over the years and start being able to think about the next steps through a more complete vision of the reality of the company and the environment.

 

Q.- Is experience no longer valuable?

A.- Yes it does, and very much so. In fact, that is why we always say that data must be an enabler for achieving the objectives set and never the end. This means that it is people, together with their experience and knowledge of the environment, that make data a true business value.

This experience is also essential to determine the most relevant variables for the statistical models, to identify errors and to be able to make the necessary adjustments to make the models more accurate. One of the most common mistakes in Data&Analytics strategies is to leave them exclusively in the hands of technologists; the business vision must permeate the entire “datification” process, from the choice of variables to the interpretation of results, including the modeling of the algorithm. Otherwise, all Machine Learning techniques will be inaccurate or simply useless, because they will not fit the real needs of the business.

On the other hand, with respect to OverlapWe know your customers’ business well, and that allows us to bring that experience from the talk to understand what they need and what steps they need to take to become real data driven companiesWe are able to combine a global strategic vision with data analytics.

 

Q.- Can an SME, or a company that is not very digitized, be a ‘data driven company’?

A.- Learning to make data-driven decisions has more to do with organizational culture than size or resources. A data-driven company can start with a small prescriptive model, or simply with root-cause analysis. When it acquires sufficient maturity and analytical culture, it will be able to tackle more complex Machine Learning projects, to finally reach prescriptive models or unsupervised models.

There are companies that often, simply to follow trends they hear about, embark on analytics projects or investments in innovative technological tools, without understanding what they really need, what is the degree of analytical maturity of their teams and how to correlate it with their business objectives.

Q.- The turning point, therefore, is to link the strategic vision and data analytics. How is this done?

A.- From my point of view, it is essential to make a correct evaluation of the four main elements that make up a Data&Analytics strategy: people, data, technology and data analytics. In addition, this whole process must be a constant evaluation over time.

I would like to stress the issue of order, because that is what really distinguishes a data-driven company. The first thing is always the employees – how they understand analytics and how they work analytically. Secondly, there is the data – as you work with it, you realize whether or not it is good enough, and whether you need new data. In turn, this is linked to technology – depending on the technology, you can go further or further; you can store more or less information. And finally, the capacity you have to analyze it in order to help in decision making.

 

Q.- And if we reverse the order, what happens if we focus first on technology?

A.- Unfortunately, this is something that happens very often. There are many companies that have large information storages and high-cost solutions with which they could make countless decisions. However, they are not exploiting these investments efficiently because the level of analytical development of their teams is far from the capacity of their tools. In other words, there is no analytical culture in the organization, nor are the processes for data exploitation well defined.

 

Q.- Since we are talking about data, how can we measure the level of analytical maturity of a company?

A.- At Overlap, in the different clients and forums in which we collaborate, we always ask a very simple question: “On a scale of 1 to 5, how important do you think data and its analysis is to help in decision-making?”. As you can imagine, the vast majority of respondents, 93% to be precise, say that this data is important or very important.

However, when you dig a little deeper, you realize that the answers do not correspond to what really happens when it comes to making decisions within the company. In HR areas, for example, only 6% to 8% of respondents make all their data-driven decisions through advanced analytics. If we go on to evaluate business areas, the percentage rises to around 30%.

At this point, the only way to get a Data &Analytics strategy back on track is to go back to basics and think about people first, working on developing their analytical skills and making sure we have the talent we need.

Q.- People’s confidence in ‘big data’ and Artificial Intelligence also has to do with how the algorithms behind it are understood?

A.- That’s right, and the more critical the decisions to be made, the more important it is for the manager to understand how the models are built and what variables the machine takes into account. That is why, in Overlap We are committed to a work methodology based on 4 steps that seek customer involvement and complete transparency so that the most technical phase is understandable to all. This is achieved by developing collaborative workflows, workshops for the democratization of data use, in-depth analysis and designing dashboards that allow continuous improvement through decisions and action plans of the decisions we are making.

 

Q.- One of the main principles of statistics is that correlation does not imply causation. Taken to the field of big data and Artificial Intelligence, how can we avoid reaching the wrong conclusions?

A.- This is another of the big keys to data analytics. We must not forget that a mathematical model will only be practical if it is based on variables that are actionable for the company. Therefore, the most important thing -even above the level of precision of the models- is that the variables you introduce into the model are very well chosen and weighted. This implies having identified the business problem or challenge very well and knowing the company and the market in depth.

Our contribution of value is in identifying the variables that impact on the result that is expected to improve in that problem or challenge of the Business. And to get to this point, it is essential to work as a team and form joint work tables between Technology and Business, where both legs work together during the process. It cannot be that a Business Intelligence or IT area is on one side and Business on the other.

“It is essential to work as a team and form joint work tables between Technology and Business. It cannot be that a Business Intelligence or IT area is on one side and Business on the other”.

 

Q.- What are the advantages of integrating people from Technology and Business in the same team?

A.- It is necessary to build a hybrid team, made up of business profiles and data specialists. The former are expected to know the Business Problems and be focused on connecting the business vision with technology.

The more technical profiles are expected to bring order to the data, work with them through statistics and Artificial Intelligence techniques and be responsible for the maintenance of the databases.

And let’s not forget that the objective is not to enter any type of data, but only those that will be truly useful. Here the work of both teams is fundamental. In addition, this way, profiles that have traditionally been technical are brought closer to the business and vice versa.

 

Q.- What profiles does a data driven company need?

A.- As I explained before, within the Data&Analytics team, we would have data scientists and data architects. And on the business side, there is a need for business analysts and even data translators who understand the mathematics and can figure out what model is needed.

These types of profiles are quite new and in high demand. Paradoxically, the most difficult thing to find are professionals who, even if they are experts in mathematics or statistics, have that capacity and knowledge of the reality of the business. We rely heavily on graduates in Business Statistics, as well as talent who have taken postgraduate courses in programming, Big Data or Business Analytics.

Tomorrow, who knows, there may not even be data scientists because the machines will program themselves. But what I am convinced of is that decision making, whether at the strategic or operational level, requires a Data Driven Culture at all organizational levels.