Building a simple lead score
When I was growing up, one of my favorite things to do was to take quizzes I found in magazines (yes, I used paper magazines growing up.)
My favorite part was trying to guess what the criteria would be for the outcome I wanted to have. I used this to become a Ravenclaw on Harry Potter quizzes.
My fascination with quizzes and assigning points to behaviors has been something I’ve been doing ever since.
Building a lead score to determine how engaged a potential customer might be reminds me so much of building those early quizzes.
To start, I think it’s helpful to consider what’s most important to a business, by using fake examples that may have come into the database.
For the fake company in this post, let’s pretend that B2B-Marketing is a SaaS B2B company that sells productivity software
primarily to Engineering and Product teams in any industry, and has two primary plans, a standard plan for companies under 1000 employees, and premium plan for companies above 1000 employees.
To see the lead score, go here
To determine what makes a high intent customer, I’ll start by determining what criteria should be included in our scoring model,
then determine how important each item is in deciding what makes an engaged customer.
- Lead 1 -
- Requested demo from form on B2B-Marketing.com
- Unsubcribed from marketing emails 3 days ago
- Job Role: VP of Product
- Company Size: 2000+ employees
- Industry - IT/Tech
- Lead 2 -
- Downloaded top of funnel asset from content syndication vendor
- Clicked on 2 emails in last 7 days
- Job Role: Director of Engineering
- Company Size: 850+ employees
- Industry - Retail
- Lead 3 -
- Downloaded bottom of funnel (high value asset) from B2B-marketing.com
- Clicked on LinkedIn ad
- Job Role: Consultant
- Company Size: null
- Industry - Finance
We’ll start with Demographic Characteristics.Since B2B-Marketing sells mainly to product and engineering teams, we’ll prioritize job roles that fall under Engineering or Product,
and we’ll also want to prioritize job roles that typically have decision making power, like Execs, Directors, VP’s and similar roles,
since these users typically research a product after the research has been done but before the final purchasing decisions are made, so their involvement
indicates that the company is ready to start using B2B-Marketing soon.
Since B2B-Marketing earns more on companies that purchase the premium plan, we’ll prioritize companies with 1000+ users as well.
Next, we’ll look at User Engagement/Behavior. This section would be the most dynamic and would need the most refinement. Typically, it would be crucial to use data from actual conversions and conversations with sales teams to determine the relative weight of different customer behaviors.
All of the behavioral/engagement characteristics are from the last 7 days since that time period is the typical time from the first engagement to purchase date
(according to my fake data). I made many of the engagement behaviors lower intent than the demographic attributes because users can have a degree of activity over
short periods of time while conducting research on a company without necessarily being ready for a purchase or sales conversation.
I considered web activity in the last 7 days to be equivalent to have a small priority, since many engaged users visit the site at least once per week.
I did not add high value web pages to the model, since I believe that content downloads and form fills are higher indications of interest, and want to ensure quality leads.
I gave social activity two categories - clicking or liking any social ad to be one category, and sharing to be another, more prioritized category, since sharing requires more effort
than liking or clicking and is an indirect endorsement of the product. I did not give any points for comments, since this hypothetical company does not conduct sentiment analysis on comments,
and thus cannot distinguish complaints from endorsements/positive comments.
I believe opening 2 or more or clicking on one or more emails per week to be worth prioritizing, and deduct points for users who unsubscribe from marketing emails during the week
since that is a sign they are less interested in engaging with us.
After conducting hypothetical analyses on actual conversions and hypothetical discussions with sales members, we determined that downloading content and filling out forms on our site or blog posts
is consistently what drives the most conversions. Thus, these have the highest priority, higher than demographic or the other engagement attributes.
I used a two tier scale to identify how high of a priority a content download was, with assets being rated either high value/bottom of funnel or top of funnel.
I am considering the “bottom” of funnel content to be the bottom of the marketing funnel, before a hand-off to sales would occur.
(Companies with a full self-service product would call this the middle of the funnel.) I would consider that to be case studies, pre-recorded demo videos,
or anything that would be used when evaluating a purchase. Assets that I considered to be top of funnel would be things designed to raise awareness of the product and/or brand
and include whitepapers, checklists, or tools/templates that provide value and intrigue potential customers.
Demo requests were given very high priority, since that score would automatically move a potential customer from a visitor to a sales accepted lead, and move them on to the sales team for more vetting.
Users who failed to completely enter their information on a form, including the demo request form would have points deducted, which would prevent them from being sales accepted.
Next, I separated users based on their scores into three categories: visitors, marketing qualified leads, and sales accepted leads. Users who did not meet many demographic characteristics but had high engagement
would remain at visitor level unless they requested demo, users with a combination of engagement and demographic characteristics would remain at the marketing qualified level unless they requested a demo.
This would give time for the marketing team to sell them over time without needing the high touch work from sales.
Ideally, this model would be used to create a predictive lead score, using a marketing automation tool like Marketo or Hubspot, which would make this process even more accurate, since they use machine learning to adjust overall scores.
This still requires a user to occasionally check and tweak as assumptions and needs for the business change.