(Originally posted on yesler.com)
Lead scoring is made up of two different kinds of information, each of which offers its own insights. Behavior score indicates sales-readiness based on a person’s activity on your website, their response to your emails, and any other activity you can measure. A demographic score is a measure of how well a prospect fits your target audience, comprising both information about the company and the individual’s role within it.

Why you should do it

Lead scoring in B2B marketing is often, unfortunately, a series of wild guesses. People think it’s a great idea, but they run into problems actually strategizing and implementing it.

With marketing automation, lead scoring should be treated as a science, so you can get an accurate indicator of both sales readiness and sales fitness. This allows you to prioritize leads effectively and measure marketing effectiveness.

Behavior score helps measure a prospect’s level of sales readiness and provides a sense of which channels provide the best prospects.

A demographic score helps measure how well you’re targeting your marketing. For example, certain channels might generate consistently under-qualified leads, something that you need to act on.

Many people lump “behavior score” and “demographic score” into one sum and call it “lead score,” but this can present problems. For example, a student might visit 20 pages using this lead scoring system and be considered an MQL. No matter how interested they are in your company and your content, some individuals will never be your customers because they will never fit the definition of your target audience.

If this is how not to do it, then how do you do it?

How to do it

To gain the best results from your lead scoring, we recommend the following.

• Keep behavior score separate from the demographic score. These are two different things and need to remain distinct. Demographic score leads to insight into where you focus your marketing efforts. Behavior score contains insight about prioritizing leads effectively for your sales team.
• Continually fine-tune. Lead scoring is not a set-it-and-forget-it action. It requires quarterly or annual re-tuning. Craft campaigns with this in mind—make them scalable (will this setup be easy to replicate?) and robust (will this setup be replicable tomorrow?). Aim to create campaigns that can be used more than once, preferably over and over.
• Correlate lead score with winning business. Run regression analyses based on the marketing activity database combined with the lead database. Sum up the pages visited, the high-value pages, the email opens, the email clicks, the form-fill-outs, the downloads, the unique web sources, and everything else applicable. Combine them all into a comprehensive CSV file that an analyst can run regressions on. Your independent variable can be a few different things, based on the sample size you have: likelihood to become closed-won (best preference), likelihood to schedule a meeting with a salesperson, and anything else down the line that is where you want to go.
• Run regressions to find the appropriate correlation. Don’t worry too much about perfect statistical significance, and bounce the results against common sense to find what you can. Turn your beta coefficients into integers and these coefficients become the incremental values.
• Review your results. These values would have been your best possible guess for lead scoring; they would have correlated as much as possible with your desired outcomes. This has statistical nuance, obviously, such as the correlation versus causation argument. Intelligent regression analysis is the real gold in lead scoring because it shows you where you should put your efforts to get the results you need.

Implement these strategies step by step. Start with an easy target and expand it slowly. The point is to increase the likelihood of a closed-win. You might have to begin with a much easier target based on the amount of data you have. Two closed-won deals in a year will not give you accurate statistics. You might have to start with measuring a lead’s likelihood to become a Sales-Accepted Lead (SAL).

Keep in mind that, with marketing automation, lead scoring is an iterative process. Because behavior score is really a proxy variable for sales readiness, it varies depending on your website, content, emails, and other interactions. The first time you set up behavior scoring, it won’t be perfect. Redoing behavior scores is extraordinarily difficult to do out of the box because using traditional methods means that you lose all of the previous lead scores and have to start over from scratch (because they’re all triggered and not easily applicable to the past behaviors for rescoring).

But there’s still hope! When you’re setting up lead scoring campaigns within your marketing automation platform, remember that you won’t get it right the first time—architect your automation to make rescoring as easy as possible, so you’ll be able to change values, run one job, and have entirely new scores factored.