You’re task with qualifying potential clients from a list of over 5,000 companies in a particular market segment and all you have is a list of company names, that’s it. For any salesperson this would be a daunting and depressing task.
The salesperson probably uses the following mental algorithm: Pick a company starting at the top (or bottom), do an online search, find the right website, read the ‘About Us’ page and ‘Press Release’ page to get an idea of whether the company would be a good fit (e.g., Right market segment, type of company, annual revenue, employee size number of locations, etc), find the right person to contact, research that person, acquire their contact info (e.g., LinkedIN) and then make the call (or send email request).
One down, 4,999 to go! If we assume the salesperson can qualify 10 a day, it would take almost 2 years (260 working days x 10 per day = 2,600 per year) to go through the list; for this one market segment.
Going back to the 5,000 companies on the list, the first task would be to determine what keywords (or phrases) are likely to point to a potential client.
Let’s assume that these 5,000 are all involved in technology in some form or another. Let’s also assume we offer a SasS application that allows these companies to develop and turnaround Request for Proposals (RFPs) quickly and more efficiently. So the goal would be to find companies that submit or respond to proposals. Not an easy task.
An NLP program could be designed to scour the ‘About Us and Press Release’ pages in search of keywords or phrases that would indicate the company is involved in bidding on deals. Such as:
[Awarded, RFP, Proposal, Request for Proposal, RFI, Request for Information, RFQ, Request for Quota, Quotation, Quote, Deal, Final Bid, Bid Winner, Government Bid, Wins, …]
An algorithm (script) can be then written to perform the following functions:
- Import Company Name
- Acquire company URL
- Locate and read ‘About Us’, ‘Company History’, ‘Who We Are’
- Search LinkedIN and read ‘Company Description’
- Combine each page read into one document (data set)
- Do a frequency keyword (phrase) analysis based on the list
- Score them according to ‘most likely’ to use our App.
- Compare the results against a test set to determine level of accuracy
Using Machine Learning to find, capture, sort and score new opportunities can be done very quickly. And by quickly I mean maybe an hour or less. Salespeople will spend less time doing administrative work that adds little value to the actual sales process and focus on other sales behavior to boost their sales.
 Nerd Speak: Convert the keywords or phrases to numbers using Bag of Words vector technique, Use TF-IDT transformation and Random Forest algorithm.