Anticipate with Big Data

Ravi Vijayaraghavan | July 31, 2010

While Internet companies such as Amazon, Google, Yahoo and others have historically leveraged Big Data to weave it into the foundation of their business, companies in other domains have, in general, not been able to follow suit.

Take the example of customer service and care. Customers interact with companies for multiple reasons through phone, web, social media, chat and other channels. They may pay their bills on the website, seek information through web chat, or they may call the company to complain about issues with its product or service. During this process, the customers reveal a lot about themselves through various conversations and actions. For example, their product preferences through a web journey, their drivers of dissatisfaction through a customer service phone call, or their intentions, when they type a search query.

The ability to piece together each of these disparate customer “votes” and preferences is a Big Data problem that, if solved thoroughly and in a scalable manner, can completely transform a consumer’s interaction with a company, making it a far more intuitive…even delightful engagement.

The goal of the Px Platform is to make this happen. The Px Platform intervenes in real-time to guide consumers to their desired outcomes. This is driven by our patent pending framework which we call Anticipate, Simplify, Learn (ASL). Here is how the ASL framework works to provide a better customer experience.

We Anticipate customer need – who are you and what is your intent?

We then Simplify engagement – Now that your intent is known, what is the best way to engage with you?

We Learn from the interaction – How do I learn “at scale” every time I engage with you and feedback the learning to improve my ability to anticipate your need better.

So how can a company Anticipate a customer’s intent?

The Px Platform goes through a two-step process to determine intent. First it tries to establish identity either directly, in the case of authenticated customers, or indirectly through phone numbers or an IP address. It then predicts intent based on customer type, history-based context, location, point in the customer journey and other such attributes.

The intent prediction models themselves are built based on large volumes of historical data. As customers interact with companies through multiple channels, the outcome of these interactions are recorded. The Big Data Platform is then used to “fuse” data from multiple channels to create a single view of the customer. This data is then used to train statistics/machine learning-based intent models. These models are then deployed on the Px Platform to predict intent for future customers.

All of this happens “behind the scenes” so we can anticipate customers’ needs and design a truly intuitive experience for them.