Earnix Blog > Analytics
A Solid Analytical Solution: Analytical Sophistication | Earnix
Jonathan Moran
18. May 2017
- Analytics
For financial services organizations as well as insurers, analytical sophistication continues to increase in importance as margins continue to shrink. With minimal margins on the majority of products and services, extracting insight from data using analytics is crucial to meeting company objectives. In this third post of a three-part series, we will look at what comprises a solid analytical solution from an analytical sophistication perspective. In our first post we looked at an overview of what makes a solid analytical solution. In our last post, we covered the Operational Excellence component, which focused on things like performance, scale, and use concurrency. Let's define what analytical sophistication entails, how to become analytically sophisticated, and example of analytical sophistication in products offered today.
I’d be happy to speak with any of the readers out there in more detail around the topics of operational excellence and analytical sophistication. In the near future, we at Earnix will summarize this blog series in a consumable document that can be downloaded and shared throughout your organization. Thanks so much for reading along – hopefully this series provided some insight into exactly how we at Earnix are thinking about the future of integrated customer analytics and the software that surrounds it.
What is Analytical Sophistication?
Analytical sophistication looks at how to not only incorporate, but also advance the analytical techniques used in enterprise software platforms. Common terms in the analytics industry are used to describe different levels of sophistication. At the fundamental level, diagnostic analytics looks at understanding what happened from a historical data perspective. For instance, what happened last year with respect to the margin decrease that we experienced? Descriptive analytics is the next stage, and looks at describing exactly why something happened. Why did these margins decrease last year- what was the root cause? The third level and probably the most widely talked about is that of predictive analytics, which allows an organization to predict or forecast what might happen in the future. For example, we predict that if we don’t change costs or expenses, margins will be the same next year, based on the data that we have. Predictive analytics is the root of emerging techniques today such as Machine Learning. And then finally, the most mature of the stages, prescriptive analytics, which details what an organization can do to change a future outcome. For example, based on the analysis we have done, we recommend reducing costs, changing the expense structure, and increasing prices in only certain regions in order to increase margins – thus prescribing a remedy. I think that if you ask any banker or insurance professional they would undoubtedly tell you that they want to move their organization up this maturity model, but the question becomes – how is this done?How to be Analytically Sophisticated?
When I think about analytics, I really believe that it is an incremental process. Become analytically advanced as an organization doesn’t happen overnight. It takes aligning people and processes with the needed supporting technology in order to achieve the goal. From a people perspective, it requires those that know the business and the data of the business, but are also familiar with analytical techniques and outcomes. I won’t put a certain title or name on these people, because I believe that at all levels of the organization people can be in place to affect analytical outcomes positively. From a process perspective, it requires setting a structure for moving from data to insight to action in motion. The data component is pretty self-explanatory – data must be sound, correct, and of good quality – easily accessible and logically organized. From an insight perspective, its all about incrementing on analytical techniques. Most organizations start with segmentation and basic model building. From here, decision workflows and optimization scenarios are often introduced. Prescriptive modeling is then experimented with, and now many organizations are investing artificial intelligence and open source technologies – and combining them with licensed insurance rating software to produce unique outcomes. From an action point of view, it requires being able to take insights and put them in place rapidly in order to affect business outcomes - whether that is getting them to other departments in the organization or serving them into social or mobile app channels. And technology has to be the enabler of all of this. If you have the people and the process but not the technology, analytical sophistication will not come forth. This technology has to be open and integrated for the best use cases to occur. For example, many advanced banks in Europe are setting up environments with data warehouses, open source distributed data platforms, advanced modeling software, ETL, and decisioning and channel engines all in a single logical instance. The result, a best in class ability to deliver individualized offers in sub second times across a variety of marketing channels.Examples of Analytical Sophistication
Earnix, and other vendors in the customer analytics area, are introducing examples of analytical sophistication that should certainly benefit customers of banks and insurance companies now and into the future. From an analytics perspective, many vendors can now help organizations use data to determine the best single offer, or a list of predetermined offers to be delivered to an end customer at an individual level (meaning each individual customers get a unique offer from a bank of offers). This has to be done very quickly, in order to be contextually relevant for the end customer. This offer can be monitored to determine if it is accepted and the reply to the offer can be sent back to the organization, such that the organization can iterate and improve upon the journey that exists between customer and organization. This is typically done through a series of processes including micro segmentation, predictive and prescriptive modeling (like Generalized Linear Modeling, Generalized Additive Models, and Gradient Boosting and Random Forest Modeling), and analytical optimization (versus priority based). These processes are supported by machine learning libraries like H20 and open source performance enhancing technologies like Spark. And all of this sophistication must be delivered in an easy to use and consume format (versus technically overloaded disparate systems), so Earnix is creating a business user layer for the manipulation and consumption of these insights as well as the ability to take action on said insights.I’d be happy to speak with any of the readers out there in more detail around the topics of operational excellence and analytical sophistication. In the near future, we at Earnix will summarize this blog series in a consumable document that can be downloaded and shared throughout your organization. Thanks so much for reading along – hopefully this series provided some insight into exactly how we at Earnix are thinking about the future of integrated customer analytics and the software that surrounds it.
Teilen article: