As we all know, Data Science is as much a science as it is an art. The key to successfully applying data science has always been ability to nail down the art part of the science. In my last couple of posts, I made the case for why we need data science for demand generation and how to leverage data science for lead identification. The questions are: (1) once you identify leads, how do you score them? And (2) do you need data science to do a superior job in leads scoring? My assertion to the second question is an emphatic yes.

Lead Qualification & Scoring

Once we identify the raw leads as I describe in my previous blog, the next step in the process is to qualify the leads and stack rank them based on certain business objectives. The objective could be to rank leads who are interested in purchasing a certain product (e.g. “… my phone broke, need a new phone…”), who are unhappy with their competitors’ product or services (e.g. “XYZ cable provider sucks!”), who have expressed positive or negative opinions about certain services (e.g. “… best camera for dynamic shots…”), or a combination of them.

Leveraging SMARTSense for Lead Scoring

The ranking algorithm we have developed is a part of our SMARTSense technology (which I wrote about in my earlier posts), and combines various algorithms from the areas of Machine Learning and Natural Language Processing. We consider the lead’s demographic data, topic contained in social conversations and more importantly, the intent expressed in the conversation. We collect demographic data from various sources to provide a comprehensive view of a lead. We dynamically figure out the topics that are trending up or down and adjust the score accordingly. Our intent detection algorithm is trained on a number of pre-defined intents. The training framework allows us to quickly train the algorithm on new intents. We expose various parameters of the algorithm to our customers to allow them to tailor the behavior of the algorithm without compromising the results (and you do not need a Ph.D. in Machine Learning to understand these parameters).

We have introduced the concept of predictive learning for scoring in order to remove the guesswork on the part of users. One of the biggest complaints users of marketing automation tools have is that there is no specific assistance provided by the tools to help them determine what to base the scoring rules on and what the scale of scoring should be. Using Predictive Analytics to look at the incoming data and adjust the scoring rules automatically takes a lot of the variance and guesswork out of the game.

We are still in the early stages, but the results are very promising. We will keep you posted as we continue to make these improvements and further delight our customers with SMARTSense.