By using website you agree to our use of cookies as described in our cookie policy. Learn More


How to Acquire, Analyze, and Act on Big Data to Predict Customer Behavior and Drive Revenue

To find the most meaningful data, organizations must move beyond how to best acquire information and focus on how to analyze and act on it.

By Scott Swartz

Today, with millions of Web sites on the Internet, creating a personalized customer-focused experience is at the forefront of every organization's business strategy. Consumers want individualized shopping and user experiences, and companies are looking to optimize their big data initiatives to provide this personalized level of service.

However, big data is both a blessing and a curse. The ability to cheaply and quickly store large amounts of information is valuable; yet, analysis tools that can make sense of the data are lacking. Algorithms and techniques that used to work well with hundreds of thousands of transactions are unusable now that the problem involves tens of millions of transactions, and the size of the data is growing exponentially.

Big data analysis must respond to real-time shifts in customer data and perform analyses of the relationships between consumers, products, pricing, promotions, and sales. For example, if sales decline, what can the company do to revitalize demand? How can the data analysis enhance promotional offers and increase profits?

To find the most meaningful data, organizations must move beyond how to best acquire information and also focus on how to analyze and act on it. If data is not actionable and cannot drive operational decisions, it is not very useful. If data can be used to assist in deciding how to best bundle services and solutions, and to determine pricing and packaging, it has value.

Increasingly, it is consumptive pricing models, not subscriptions, that will influence consumer behavior and service usage. Consumptive pricing will help drive customer and partner behavior, but to do this, companies must harness big data from many sources including in the cloud as well as in mobile and virtualized environments. Turning data into information will ultimately help organizations increase revenues, enhance growth, and reduce churn.

User data can typically be broken down into two distinct types: transactional and sub-transactional data. User purchases or transactional data isn't typically thought of as big data, and although such data is valuable from a monetary standpoint, it only tells part of a story. However, when you add in sub-transactional data or clickstream-based data, you get a story about the user's behavior and intentions that is far clearer and more predictive.

If you know that I bought a book from for $14.95, you only have limited view of the user. When you understand what other books I bought prior to this purchase; what reviews I read; what time of day, month, and year I make purchases; from which type of device I shop; and how often I look at other items prior to purchase, and you compare that information with other purchasers who have similar shopping paths, you can predict my future behavior. Amazon commercialized this concept of item-based collaborative filtering to generate suggestions for shoppers on its site. The company has also been able to predict purchasers' behavior based on their visits from tablets or PCs.

As developers of applications today, you also need to consider how you architect your solution to track user behavior or engagement with your solution. A simpler way to start may even be to leverage the crop of new big-data customer engagement applications popping up to take advantage of insatiable demand to know more. By simply including JavaScript code snippets, you can create an application to track many details about the user and his or her click-stream behaviors. You can even configure these tools to provide real-time, in-app messaging or offers to customers based on their specific behavior patterns.

Having the ability to track a user's behavior and offer suggestions or offers is only part of the solution. Being able to drive dynamic offers and then monetize those offers immediately is where you can act on this rich data set to drive more revenue. To move this from the realm of science fiction to fact, think of airlines that change their prices based on the sales trajectory of a specific plane's seats over a specific time period. This constant re-pricing based on purchase data is called yield management. With today's off-the-shelf technology, you can track your users' behavior on a more granular basis than simple purchase and yield timing. Then, by adding in multiple offers and a flexible commerce engine you can bring revenue to light that simply wasn't available before.

Improve Customer Satisfaction

However, just as important as selling your customers more products and services may be, when you understand their actions and activities in-app, you can address an even bigger issue for recurring-revenue-based business: churn. Customers lost to churn need to be acquired, and in today's online world, keeping customer acquisition costs (CAC) at less than their annual recurring revenue (ARR) is important.

Simply looking at the transactional data may show the value of a purchase or a customer, but by looking at the sub-transactional big-data, you can learn when customers begin to use your solutions less frequently, when they are hitting specific computational or functional walls, or even when they are about to make specific mistakes using your application. By fine-tuning your user experience, offering functional alternatives or stepping in to prevent problems, you can drive a far more effective customer success model and prevent dissatisfaction before it arises. Although it might seem like a privacy issue or user stalking, real-time data analysis leads to far more engaged customers.

Subscriptions are Dumb; Consumption is Fat

Many cloud and SaaS-based business models focus on the beauty of subscriptions. This is often reinforced by venture capitalists who preach the mantra that recurring revenue is an annuity stream than can carry astronomical valuations for practitioners. However, relying on subscriptions can be dangerous. Subscriptions tell only part of the story. They represent dumb transactional data that is inherently limited in providing a holistic understanding of your customer. Understanding something about usage and even building pricing models that take into account a user's clicks, downloads and views enables you to scale pricing in a far more flexible manner.

In our work at MetraTech, we see far more businesses moving beyond simple subscriptions to consumptive-based business models. Consumptive models are rich with fat, big user-data that can be mined and leveraged for additional revenue.

The future belongs to those who know their customers not just on a superficial, persona-based level. Winning businesses will leverage how their users speak through their purchase and usage patterns, and those businesses will respond with pricing and offers based on real-time awareness.

What Future Commerce Requires

A3 -- acquiring, analyzing, and acting on big data -- isn't as complicated as it sounds. Begin by introducing big-data-activity-based capture into your online solutions. Track user behavior and provide offers and options based on those actions. Then introduce a modern commerce-and-compensation engine to act upon and monetize your insights. You'll be able to model and test nearly any pricing model and quickly respond to your users, thereby driving revenue in ways that were never possible before.

Scott Swartz is the founder and CEO of MetraTech. You can contact the author at

TDWI Membership

Get immediate access to training discounts, video library, BI Teams, Skills, Budget Report, and more

Individual, Student, and Team memberships available.