By Andrés Florez Llano
The Path to Analytical Maturity
Let’s talk a little about Business Intelligence, Business Analytics, and Big Data, and their value on the journey toward achieving analytical maturity. For this, we must introduce the concept of the Analytical Route Map. The Analytical Route Map is the path (redundantly speaking) that leads us to be able to use analytics in a company. These are the series of steps to take in order to achieve that much-desired maturity. They include all the stages, techniques, and processes that we go through and use to handle both.
It’s important to emphasize that this is not a waterfall process. They can unfold in parallel with the processes of BI, BA, and Data Governance. It is also not necessary to have 100% polished Data Governance because data enrichment is continuous; that is, we will never have absolute governance over data. It’s not necessary to have 100 variables to generate models—15 are enough to start generating value.
Business Intelligence
BI, or Business Intelligence, can be framed within the initial stages of analytics in companies. The concept became popular about 5 or 6 years ago. Large tech companies aimed to democratize information presentation. Many large tech companies wanted tools that could connect to these information systems and provide a functional business user with the ability to see deviations, groupings, and trends visually. These tools simplified the high volumes of raw data through aggregation processes. Over time, the need to manage information using cubes and business structures like snowflakes or star schemas grew, complementing the process with navigability in business hierarchies.
In the Analytical Route Map, what we can frame within BI starts with access to multiple sources of raw information, moves through Ad-Hoc reporting and automation of reports, and ends with dashboards—those famous control panels—where people no longer need to request information from IT/Data Governance/Systems for the views they need. Instead, users can directly connect to the database, create their own cubes, and share near-real-time information with the company.
Dashboarding techniques and BI go hand in hand because the goal in the initial stages of the Analytical Route Map is to answer descriptive questions, such as: “What is happening in the stores?” or “What is happening with sales?” and gain control over events that are happening in real-time, telling the story of what happened and why it happened. BI through Dashboarding provides clear answers about what is happening in a descriptive manner, supported by clear data visualization techniques.
Let’s use a simple example: we are in the car, and the dashboard is the temperature or fuel gauge in the car’s panel. If the gauge starts turning red, you need to take action. If the fuel gauge shows you’re running out of gas, you can react and head to a station. You can respond to a car overheating or running out of fuel because your dashboard gives you real-time tracking of everything that happens in the car. This same logic applies to businesses.
To summarize: in the first block of our Analytical Route Map, we try to provide answers about what is happening in real-time, automatically, periodically, and in a standard way, using Business Intelligence.
Business Analytics
When we advance along the Route Map, we move from Business Intelligence to Business Analytics. This is where the concepts of Data Science and Data Mining come in. At this stage, we are no longer just answering what is happening—we can also answer what could happen. In other words, we move from descriptive to prescriptive analytics. In this second stage, we apply other techniques to model historical events and predict what could happen.
Just as the dashboard was a key milestone in BI, models were a major milestone in BA. Having a model deployed in production is crucial.
What are these models?
- Profiling models
- Customer segmentation models
- Recommendation models
- Forecasting models
A segmentation model is one used by a company to divide customers into categories that make sense, using precise information, not inference. Recommendation models are used for things like Instagram ads showing you products you like, which gets better as you interact with them more.
Here, we begin to improve data treatment techniques by using statistical processes, Big Data, and Machine Learning algorithms. It’s important to understand that the goal of all these stages, techniques, and processes is to recognize patterns in the data and react to them. It’s not about machine learning techniques being a mere improvement of classic models. The goal is always to find the right models and techniques that truly create value.
Big data
Big Data was an evolutionary process in computational power and data storage capabilities. Big Data introduces a new paradigm in traditional modeling processes. With Big Data, you no longer face a lack of information, and you don’t need hypothesis-based or simulated data techniques. Now you have plenty of information. For example, when doing customer segmentation, using Big Data means you’re not inferring that Group A and B’s consumption equals an inaccurate average. Instead, you’ll know exactly how much each group consumes in a much more granular and complex segmentation—or as simple or complex as needed.
The retail sector understands perfectly the importance of Big Data. They know they must focus their vision, management, and company on the customer. They understand the customer from their profile, create models, understand their maturation process, their Lifetime Value, as well as their consumption patterns. The retail sector stands out in using analytics because it centers its processes on the customer and then applies modeling techniques from there.
When it comes to Business Intelligence and Business Analytics, it’s important to emphasize that it’s not about choosing just one. Each process becomes more relevant as the company advances along the Analytical Route Map, as they tackle different issues and create value at different points. With BI, you get clear, actionable answers in the early stages of your Route Map. Then, in the second stage, we have BA with more complex techniques and concepts such as Data Science, statistical processes, Big Data, and Machine Learning, which provide answers to what could happen. Finally, we reach the third stage, where we look at what needs to be changed to make what we want happen, to meet objectives, and reach analytical maturity.
