We present a comprehensive lead scoring system that will help take guess work out and help your sales machinery focus on the right leads. What makes this a comprehensive system is the fact that it takes into account thousands of data points across the following systems.
- CRM – Salesforce
- Marketing Automation – Marketo, Pardot
- Social – Twitter, Facebook, LinkedIn
- Finance – Public Finance information from Yahooo
Here are the implicit and explicit parameters that we’ll collect information on:
|Customer Demographics||Page views||Twitter sentiment||Revenue|
|Customer Region||Number of searches||Facebook sentiment||Market capitalization|
|Industry||Number of downloads||LinkedIn to mine open jobs||Year over Year growth|
|Lead Job Title||Number of emails opened||Profitability|
|Lead Department||Number of email clicks|
|Landing page visits|
|Number of case interactions|
Before applying a classification algorithm, we have to first run sentiment analysis on Social data. Jeffrey Breen has these slides explaining how to do sentiment analysis on Twitter. The high level overview is to calculate a sentiment score for each tweet. The main ingredient to calculate such a sentiment score is to create an “opinion lexicon” in English. Fortunately, Bing Liu, Minqing Hu and Junsheng Cheng have created just this and is available here.
We then need a function to calculate the sentiment score, and for the purpose of this prototype, we’ll use R function created by Jeffrey Breen.