Monday Oct 29, 2012 by Aki Balogh - Sr. Manager, Customer Insight at Calpont
In the world of Big Data, there has been a flurry of investments in the infrastructure layer. But activity on the analytic application layer -- i.e. using Big Data tools to solve real-world problems – has been a lot quieter.
What new business models can be built with Big Data? Having been a management consultant, a VC Associate and an entrepreneur in the analytics space over the last few years, I've investigated this question from several different angles.
The real-world applicability of Big Data haunted Xconomy's recent Big Data conference as well. Distinguished speakers such as Brad Feld and Chris Lynch held diverse opinions on whether or not Big Data is merely a marketing label or whether it is actually a burgeoning new industry that enables startups and enterprises to generate revenue in new ways.
Part of the confusion stems from the use of the term 'Big Data'. The use of Big Data implicitly assumes the presence of analytics. If a small, simple, slow-moving dataset provides you an amount of analytical value, as the reasoning goes, then the analytical value of Big Data will be much greater.
However, virtually every successful company today is already data-driven and analytical. If you don't use data to make decisions, you don't know what's important and what's not. Companies that don't make decisions through some type of analysis never scale to significant size.
In other words, if you're a company in industry X, you grow when you know how to run your business well. Let's denote this as point (1). This seems like an obvious point but often gets lost in the discussion. For example, Walmart executes well in various functions related to retail (site selection, buying, inventory management, etc.) which drives financial success.
On the other hand, the presence or use of Business Intelligence is an entirely separate discussion. In order to manage their data-driven decision-making, companies have BI functions that help turn data into information for decision-makers. Let's name BI point (2).
When you work with Big Data, you enhance the volume, variety and velocity of your data. So, Big Data can enhance either (1) or (2) above. However, doing (2) really well doesn't mean that you're automatically doing (1) really well and vice versa. Walmart is not a great retailer because of their BI function; they're a great retailer and, among other strengths, have a great BI function enabled by a strong IT infrastructure.
How does all of this tie to new business model creation? Analytics, by themselves, can't create a new business model, just like a great BI function can't make up for poor business decision-making. But, you can create a new business model that is delivered through or enabled by analytics.
Restated differently, serving analytics without a specific business context is a hard sell, because you're not 'just' serving analytics; you're also serving the business judgment that you've baked into the analytics as well. (How does the COO of Pepsi know that your approach to running inventory management is better than hers?)
So, then, what do you use Big Data for? A few suggestions:
1) You can create new user experiences with your products. For example, you can produce a better mobile app because you're analyzing data on how customers prefer to use the product. Or, your e-commerce website captures the popularity of items and automatically shows them on a 'Featured Items' page.
2) You can predict new events that might occur. So, if you're already tracking retail inventory shrinkage, you could get a better result if you use larger or more varied datasets to detect behaviors that typically suggest shrinkage. (Of course, taking that shrinkage-predicting algorithm to another problem -- say, logistical planning -- would probably be useless.)
3) You can identify latent connections in the data. Thus, when new data is added, you can identify characteristics of that data based on the dataset you already have. (Think LinkedIn's suggestion features.)
This is not meant to be an exhaustive list. The point is that creating value through Big Data really stems from industry expertise. Are you very good at retail management? Now, instead of providing strategy consulting services to retailers, you can provide data-driven consulting through analytics.
Well, you ask, what about the analytics application companies that are in existence today? Successful analytic application companies are typically not in the business of 'just' providing generic analytics... they're vertical applications built on horizontal Big Data platforms that solve very specific problems.
So, the formula for building a great Big Data startup is effectively similar to other forms of technology entrepreneurship. Find a significant market opportunity and a business problem. Then, figure out a way to alleviate this problem by working on large and/or diverse datasets, potentially in a real-time fashion. Then, build a product, get out there and sell!