Successful Predictive Lead Scoring Requires Rich Data
- Written by Brian Anderson, Associate Editor
- Published in Industry Insights
The report shows that 58% of the B2B marketers surveyed state that they want predictive models for their lead scoring programs. But incomplete or inconsistent data (59%) and lack of knowledge on what attributes indicate buying behavior (44%). Lattice Engines CMO Brian Kardon stated in a recent webinar that it’s not that marketers are doing the wrong things; it’s the data being collected that’s coming in under par.
Kardon added: “Companies have all these tools — marketing automation, buyer personas and lead nurturing — in place, but despite all that the results are horrible. After all of this sweat and blood to implement the right tools and strategies, 94% of their marketing-qualified leads will never close.”
The basic data that’s collected from leads — name, title, annual revenue and various others — is too generic for marketers to build a predictive model. Predictive lead scoring requires marketers to go deeper into the data to find the common attributes that indicate a likelihood to purchase.
“The data that we are taking advantage of is only the tip of the iceberg,” said Kardon. “While the data that marketing automation provides you gives you good insight on your leads, there are hundreds of different attributes that are hiding in plain sight that can be incredibly predictive and much more behavioral.”
Having a predictive analytic solution with a “recommendation” feature is helpful when your data is either unreliable or overwhelming, said Amnon Mishor, co-founder and VP of Products at Leadspace.
“For example, when you go on music streaming sites like Pandora, you don’t provide your entire iTunes library in order for the program to do its job,” Mishor said. “What you normally do is provide one piece of information — like music genre or a specific singer — that you know you want, and the program provides a wide range of music based off your criteria.”
Although having a firm understanding of your company’s buying persona has its benefits, it’s not considered a necessity to some marketers in order to begin predicting where your next sale will close. According to Doug Camplejohn, Founder and CEO of Fliptop, predictive lead scoring is possible as long as you have the proper technology in place.
“The idea that a company needs to manually define what their trigger is and the things that make up their model is not really the case when you are using a modern solution that automates the process and have a large quantity of data,” said Camplejohn in an interview with Demand Gen Report. “If you have the technology, let the math tell you what can be used to predict potential new customers”
Improving Predictive Capabilities
Although the task of implementing predictive capabilities into your lead scoring strategy may seem daunting, today’s advancements in technology allow marketers to become predictive with very minimal effort. Various solutions offer a wide array of options for marketers.
“Marketing operations teams have started incorporating predictive lead scoring systems that range from black box approaches to more “plug and play” SaaS based solutions,” said David Lewis, founder and CEO of DemandGen International. “Although the range of predictive scoring solutions varies widely in complexity and sophistication, it’s very exciting helping our clients evaluate and leveraging solutions from vendors.”
Progressive marketers are turning to third parties to help them build their predictive lead scoring models.
“Building predictive models ia really difficult to do,” said Nate Gemberling, Account Executive at Infer. “For example, you’re not going to go build your own search algorithm for your company — you’ll go to Google or Bing for that. It’s the same thing with predictive lead scoring; you want to partner with vendors who already have the processes and algorithms in place.”
Partnering with a predictive solutions vendor has minimal impact on the company’s business model and can provide a fresh perspective.
“Sometimes there are particular data points that make common sense to be used when creating a predictive model, but there are many different data points that may not be seen with the human eye,” said Dr. Jacob Shama, Co-Founder and CEO of Mintigo. “Let the mathematical programs do their job, and they will help find the most relevant data points that you need for a specific customer.”