Analytical problem solving with Carlos

Carlos Pinheiro, PhD, Principal Analytical Training Consultant, SAS

Carlos Pinheiro is talented. One afternoon Carlos stood in front of a group of marketers including myself, and shared social network analysis using our data.  Yes, the analysis took talent, but the real brilliance was that Carlos presented the information in a way where we all left the meeting with an understanding of his analysis, and actions we could take to improve our business.  I don’t understand a lot of things…the movie Inception, the French language or how ships are made in a bottle, but I now have a basic understanding of social network analysis thanks to Carlos.

Carlos has a new course, Analytical Approaches to Solving Problems in Communications and Media, that will be making its debut at Analytics Experience 2016 in Las Vegas. I recently caught up with Carlos to ask few questions about his insight into trends in the industry and his new course.

Q:  As a data scientist, what kinds of businesses problems have you seen in your work in communications and media?
A:  While there are various problems we can face to in the industry, there are also plenty of opportunities. Traditional business problems like preventing churn and boosting cross-sell and up-sell opportunities continue to be points of focus. Companies have ongoing processes in place for these actions, with fraud as a recurrent topic. Businesses can save tens of millions of dollars by detecting fraud early in both usage and subscription scenarios. Ultimately, it’s a highly competitive market and companies in this industry need to understand the complexity of their customers’ behavior.

Q:  How does a solid analytical approach help businesses better understand their customers?
A: Everything customers do in terms of consuming products and services leave a trail. The way we follow this trail is by analyzing the data in relation to it. A deep exploration analysis upon customer usage data can reveal particular consuming behaviors. Unsupervised models – such as clustering – allow companies to envision similar groups of customers in terms of products and services they’re consuming. This knowledge is fundamental to both better designed bundles and packages as well as customer-driven campaigns. Based on historical business events, companies are able to build supervised models – such as regression, decision trees and neural networks  – for an extensive variety of cases. Churn, cross-sell, up-sell, bad debt, insolvency, fraud, acquisition, all of these can be predicted – with a certain level of accuracy – helping companies to prevent or anticipate business events.

In addition to the traditional unsupervised and supervised models, a solid analytical approach also includes social network analysis, path analysis, text mining and optimization models. These distinguish techniques help companies understand not only customer behavior – the relationships between the customers and their products and services – but also the relationships among them. How customers are related to each other and how these relationships affect their decisions can reveal customers’ influence upon particular business events. The way customers interact with companies over time, from the subscription to the consuming phase, can expose a common path that might be applied to other customers, allowing several business actions to be put in place in order to boost operational processes.

What are customers saying about your company and its products and services? All this unstructured data can be tracked and analyzed to recognize possible patterns and then to create or improve business actions. Even when we don’t see a problem, things can always work better. Optimization models can help companies improve their operational processes, increase quality, reduce casualties, or diminish faults. At the end of the day it saves money.

Q:  What is the customer life cycle?
A:  Whether it's telecommunications and communications, or media and digital content, we have basically three customer phases. The first one is when we need to acquire them. We might need to sell them a SIM card, a mobile phone, perhaps a broadband, or a TV channel, or maybe a specific digital content. Efforts associated with these actions have a cost. Analytics can help companies to better target their audience, increase the response rate, and save money. Once we gain customers, we need to make them happy, fulfill their needs, and keep them engaged. A solid analytic approach as you asked can dramatically help with this, allowing companies to better understand their customers’ behavior, and adjust actions accordingly.

Finally, and unfortunately, we have the last stage of the customer’s life cycle. It is when the relationship ends. We can see this event in two distinct ways, the voluntary and the involuntary event. For the second one we don’t have too much to do. People move (between cities, states or countries), they lose their jobs, or their health diminishes. There is nothing we can to about it.

However, for the voluntary churn, we have very much to do indeed. Most of the customers leave a clear trail telling companies they are about to make churn. Companies need to follow these trails and prevent customers from leaving too soon. A mix of unsupervised and supervised models can powerfully assist companies in this challenge, by revealing the different churn behaviors and capturing their patterns, making it possible to predict them afterwards. Specifically in communications and media, where customers commonly relate to each other, social network analysis, path analysis and text mining can help companies understand possible influences.  The domino effect isn’t uncommon in churn, where one unhappy customer can influence a large number of their peers to leave as well.

Q:  Who is your new BKS course designed for?
A:  This course is mainly focused towards business, marketing and data analysts. Data scientists, mathematicians, statisticians, data miners and data modelers can also benefit from the concepts we cover. Even people from different industries outside of communications and media can be benefit by the problem solving approach presented.

Q:  What kind of insight can be gained from the analytical models that you cover in your course?
A:  More than anything else is the analytical mindset. For any problem, or any opportunity, there is a set of different solutions. Some are better than others. Some are easier than others. Some are faster than others. Some are cheaper than others. With the proper fundamental knowledge we can decide the best one for each particular case. There is no secret ingredient. We must try, and try again, and try as many times as we can in order to find the optimal solution for a particular problem or opportunity. We need to then reassess that solution from time-to-time and see if we can improve it. This is a continuous cycle. As we address a problem or opportunity, we map all the data in relation to it. Then we build a model to solve it. As we deploy this model into production, we hope to change that business scenario.  If you succeed, the scenario changes and therefore the data that describes it. Through this change, our model becomes obsolete and we need to rebuild it.

Q:  What future trends are you anticipating in the field of communications and how will analytics play a part?
A:   Telecommunication companies are now offering content. Media companies are now offering communications. Different industries characteristics are melting together. I believe the complexity of products and services will increase. A combination of communications networks, devices, content, media, everything will be somehow connected to each other. Customers will have increased numbers of customized choices. Understanding all these different behaviors and patterns will be a key for companies to thrive. And for sure, analytics will play a fundamental role on this, revealing those behaviors and patterns along the way




About Author

Mike Janes

Senior Product Marketing Manager

Hi, I'm Mike Janes. I'm going on my 16th year here at SAS. I worked as a project manager and technical support engineer before joining the Education and Training division in my current role as a product marketing manager. I owe a lot to SAS Education and the numerous courses that I've taken over the years. I went from being completely unfamiliar with the SAS language to now manipulating data and creating reports using SAS.

Comments are closed.

Back to Top