Leveraging Data Science for Lead Generation – Part 2

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 post, I made the case why we need data science for demand generation. Today I will discuss what it takes for finding good leads on social channels using data science.

Lead Identification

An approach that many of our competitors have been using for lead identification is to perform search on the social data and then rank the result based on certain criteria. The primary problem with this approach is the way they form the search queries. That is the keywords that they use to perform the search. Usually, they use the set of well-defined keywords that explains their customers business.

Unfortunately, this approach results in limited success. Conversations in social media are inherently ephemeral and highly influenced by “immediate” events. The conversation that is popular today may not be popular tomorrow i.e. today’s search criteria may not be relevant tomorrow. For example, when, on March 15th 2014, the news came out that Jim Kelly, former QB of Buffalo Bills, had a recurrence of cancer, the social conversations under the broader topic “NFL” or “Cancer” were completely dominated by that news. A smart system would understand the variation in the conversation and accordingly adjust (include or exclude) the newly found search criteria. This implies that we need a scalable mechanism to determine search criteria that takes into account the dynamic nature of social conversations.

To solve the above problem, we explored various algorithms in Natural Language Processing, Information Theory, Statistical Inference, and Machine Learning. We also explored various ways of determining trending topics at a micro level and applying decay factors. The technology that we built borrows concepts from these areas and combines several algorithms to address the problem of dynamically discovering keywords given a broader topic. We call it SMARTSense.

SMARTSense from NextPrinciples

Our SMARTSense technology captures the changing trend in conversations under a given topic (micro-trend) and auto-adjusts to listening to the conversations that are relevant in the present context. It also introduces two critical notions, which we call “relevancy” and “reach” to further refine the lead identification. These two notions will provide our customers the ability to fine-tune the algorithm to suit their business needs. The Relevancy Dial provides the ability to focus on the conversations that are highly related to their topic of interest. The Reach dial provides the ability to control how broad an audience they want to listen to.

Early results have shown great results using data science based SMARTSense which the marketing team at NextPrinciples will share in the near future.

Have you leveraged data science for lead generation? I would love to hear from you.