Data: a tool to make companies grow in Latin America
“I love that analytics can find a way into any passions a person might have. For example, I love soccer, and hadn’t really thought about it from an analytics viewpoint. And then, when I actually did, I was amazed by all the correlations and patterns I could find. I never looked at soccer the same way again. Germany won the 2014 Soccer World Cup and they explained that analytics and Big Data were a big part of this result”. When Francisco Calero, Analytics and AI Consultant at Kin, speaks, you can feel his excitement. His eyes light up with so many ideas and examples he takes from real life customers at Kin Analytics in Quito, Ecuador.
Francisco recalls having always been a gifted student, especially in Mathematics. But he never saw himself as a scientist. “I’m very practical, and not really inclined towards theory. I ended up deciding to study Industrial Engineering because of the way it optimizes processes”. As an undergraduate he decided to have an exchange experience at the University of Illinois at Urbana-Champaign. There, he enrolled in a number of courses oriented towards graduate students. And he discovered Big Data and Statistical Analysis. From there, everything spiralled into shape: he became part of an MIT project that used research and analytics to identify and characterize freight transportation in Quito and impact positively on the local public policies in place.
This turn of events led to others, and he finally ended up meeting Andrés Pérez, founder of Kin Analytics. This was enlightening for Francisco, as he had realized that a career in consulting was something he was suited for. He’s been there for over two years. “It’s a place for me to research, be curious. Andrés believes that consulting is a way to develop new products. So I sit with my customers and think alongside them about the challenges they are facing, and come up with an analytical model. As we grow we find ourselves doing more and more business continuity work”.
One of the industries that Francisco believes are very interesting from a data point of view is Logistics. What’s revealing for him is that Latin America is often studied as a high complexity case, but mostly from the outside. In Latin America he believes that only research centers actually conduct investigations on the matter. “But in the private sector it’s pretty much old school. This doesn’t mean that there is no existing data: most companies have GPS tracking because of safety concerns, but don’t really use it much outside of it. This means that there is an enormous potential to optimize outcomes. And sometimes the private sector executives don’t really understand the value of data, because they have done things in a particular way all their life and see no reason to change it. And at Kin we’ve come to see that most of the times business people pay attention to analytics once their market starts discussing it, but the risk there is that their competitors might have already discovered it”.
Other industries Francisco identifies as particularly interesting are finance and retail. He finds the latter, particularly, as a data factory with multiple opportunities to get to know their customers and their contact points. This means that benefits can be reaped earlier. But at this moment this industry in Latin America is not entirely transformed, although there are early adopters.
The financial industry, in Francisco’s view, is less developed in the continent. For example, credit scoring is less used in Latin America than in other places. In the case of Ecuador, healthcare had major improvement opportunities. This became evident when Covid hit “because they weren’t unifying their information, they had no integrated data on critical areas such as the number of available hospital beds”. Kin worked intensively on this.
Francisco explains that for the latin american private sector, at the beginning, there’s no confidence in data. The explanations for this are varied, such as the distance between academia and industry, the curiosity of middle management, the education of business people.
“Something I get a lot from companies that are considering becoming data driven is that they are not going to act on data until they are 100% sure that the data is real. And I understand, they don’t want to act on something they are not completely sure it’s a fact. I think this is related to the culture of error in Latin America: we don’t value the risks that people take, but we do value certainties. We prefer people that play it safe and end up punishing the mistakes made along the way instead of seeing it as a part of the process. If you don’t make mistakes, you are probably not moving forward. Being wrong isn’t the end of the world, you just need to readjust and carry on. I think it’s preferable to make decisions based on the wrong data than on no data, but people feel differently sometimes and would rather trust their gut. Sometimes information can be biased, but not totally untrue”.
This, as Francisco says, might be related to the confusion some of the discipline’s terms cause. “For example, AI and analytics are often confused. But AI implies the algorithm at play can make decisions on its own as well as learn. If it needs someone to manage it, then it’s not AI. And analytics refer to the way we use the information that was obtained and turn it into usable data that is relevant to the problems or issues we have to face. Having buzzwords that create interest is good, but it can also be confusing”. To illustrate, Francisco provides an example: “Right now, the traffic control cameras in Quito are also detecting whether or not there are agglomerations of people. And that shows the joint work of AI and analytics”.
This takes on a new relevance when we consider the ethical aspect of the discipline. “Shortcuts are never good advisors. If I sell something that sounds fancy to a customer but ends up being quite different from it, I’m at fault. I shouldn’t tell my customers that I’m going to provide them with a machine learning service when I’m actually just going to collect information. But there are more ethical challenges that the discipline has to face: for example, we sometimes run into models that are biased.
This happens a lot in the financial industry, mostly in credit scoring, where gender plays a major role because historically significantly more males ask for credit and repay it even though women do it to, but in sensitively smaller numbers. Does this mean that we have to build a model where gender determines whether a person receives a credit or not? They are historically biased, and sometimes other variables are important as well, such as zip code. For us this is morally conflictive, because we don’t want to deepen the biases. The real challenge is making the algorithms not learn from these perspectives. And one way to stay on the ethical side of things is being close to academia and have their input on whether we are doing what’s right for society, for example data protection”.
According to Francisco, the discipline is currently limited by two big issues: the budgets that companies or organizations allocate to data, and the availability of information (the construction of databases, with the monetary and temporal costs involved). And from an organizational point of view, another limitation is the lack of historical data.
“Organizations and companies often don’t keep their historical records, or their commercial and marketing initiatives, in order to go back and reflect on what’s been done. And this, to us, impacts directly on our ability to define more complex models. This is a significant roadblock for us, and there seems to be a cultural force behind it that makes it hard for people to overcome it. I think that if we don’t see record keeping as a part of the life of the organization, our results will be as poor as our data. But then again, there is always something we can do with the information we have. For instance, what companies do keep is the invoice and billing information because they are usually legally required to do so. And this is a good way to start figuring out the customer’s shopping habits. But the bottom line is that if we start collecting information now, in six months or a year we’ll have enough information to start building a model that works for us. But this is really up to the organization or company’s leader: if they aren’t on the same page, it will likely not work”.
For Francisco, the time to move forward with data is now.
“It’s better to do it now than when your competition has already started and you’re falling behind. Sometimes customers ask us if we have used whatever model we’re applying before, and I tell them that if I have already applied it, it means someone else is ahead, so it’s actually a good thing that it hasn’t been done before.”
Francisco wholeheartedly believes that the sky’s the limit. This world is full of information and access is simpler than ever. “So go on, explore, search, learn on your own. We all need to be constantly learning because the world constantly changes”.
We at Arion couldn’t agree more. 2020 is a black swan that makes us reflect and think not only of what we’ve left behind, but of everything that is coming, in a very holistic outlook. Simply put, we believe in transformation. This new reality means new opportunities, new ways of doing things and definitely new approaches to existing processes. This is where digital transformation as a whole comes into the picture and impacts our life in multiple and powerful ways, whether it be being data driven to make the correct decisions, or deciding to digitize what’s always been doing presentially, or creating new, entirely digital ways of doing things. That’s where the value is now, and we feel confident we can deliver it.
