“Who moved my cheese” is the title of the famous management book written by Spencer Johnson. It treats the fact that some of our current and beloved business models will fade out sooner or later and that you should realise and embrace the point in time when a paradigm shift is about to hit your business model.
As an example, think of retail banks who sold their products primarily through a network of branches. Today, the number of these points of sale is dropping like a rock. Obviously, this is because an ever increasing amount of people finds it way more attractive to manage their finances through user friendly apps while sitting on the couch.
Data Science is frequently classified as being a disruptive collection of methods and a new paradigm of data driven business models for many (if not all) industries. Especially industries in which margin pressure is high, like the retail and e-commerce business, or those who show a high affinity to collecting data, like online advertising, spearhead this movement.
Other industries like the Financial Industry for example seem like perfect candidates to exploit the power of data science. They, however, appear to be among the laggards in adopting these new possibilities. This is surprising for two reasons:
First, the upside of leveraging the potential of data science and analytics and developing data driven business models is not only a measure to increase internal process efficiency but especially to attract customers and maintain a sustainable business. Second, the risk of a “sit tight and wait” strategy is truly suicidal. Establishing a data driven business culture cannot be done over night and needs time for people training and development, letting aside the effort and time needed to choose and set up the systems and infrastructure.
Recall how Google disrupted the search industry. Yahoo, Lycos and all these almost forgotten dinosaurs could never catch up over come even close to Google’s success after they had been disrupted. Hence, from the day on when some financial institutions start to seriously follow a completed digital vision, the remaining air will become thin for the laggards in the financial system.
So, which road blocks are holding the financial world back?
- it’s hard to determine a ROI figure before first results are available
- data is still hard to be made available on a continuous basis
- regulation limits the number of possibilities in combining data
1. It’s hard to determine a ROI figure before first results are available
It’s true that you can not tell how well you might be able to forecast e.g. customer behaviour based on your data before you haven’t tried it. The first stages of a data science project will contain some experimentation and trial-and-error elements. This might feel undesired but is an inherent characteristic of many data science uses cases. However, this is not an excuse to not do it. Rather, you should have some R&D budget available for these sort of projects and to collect experience points. Second, if you don’t try to boil the ocean but rather laser focus on some promising use cases that have been selected following the DIFA framework, chances are high you will strike a rich vein.
2. Data is still hard to be made available on a continuous basis
Usually, in discussions about which data might influence an outcome, the sky is the limit. Including all data sources into an analysis that might influence the outcome is like building the Chinese Wall. At this point it is wise not to try to boil the ocean.
There is a very neat way to prioritise data sources, as shown in the data prioritisation matrix below. In brief, you want to start with the bread & butter quadrant and only extend your data basis beyond this, if it’s absolutely necessary. In many cases, it’s not.
3. Regulation limits the number of possibilities in combining data
The finance industry and regulation have had a tough time over the recent years. Hence, the impact on regulation regarding data and analytics issues is perceived more severe than it actually is. Of course, there are clear lines that may not be crossed and reputation risk should be part of the discussion.
The critical point are personalised data. And here is the good news: Analyses can almost always be carried out without personalised data. Since the goal of most analyses is to discover patterns, clusters and the like, a surname or phone number doesn’t really help here. The bottom line is to stay in line with regulation but at the same time not to be overly cautious about it.
Vita: Alexander Beck holds a masters degree in physics and a PhD in economics. He has worked as a data scientist for a quant hedge fund and has spent several years as a data science and pre-sales consultant. Today, he leads a data science team at a startup company in the financial industry.