In its brief history, predictive scoring has been mostly applied to leads and opportunities. Predictive models are built to help answer questions like "Which leads are most likely to convert?" or "Which opportunities should my reps spend the bulk of their time on?"
With the rise of account-based marketing and associated outbound sales processes, the focus of predictive scoring has shifted a bit. While models are still being built to score leads and opportunities, the primary focus of predictive scoring within an ABM approach is on the account level. "Tell me which accounts will have the highest lifetime value for my business."
Here's a quick video outlining how Datanyze solves this problem with our own Predictive Account Scoring solution.
The following post outlines the basics of predictive account scoring and provides some tips for how to use Datanyze Predict within an account-based approach. If you're already familiar with the basics, feel free to jump down to "Our Approach". Comments and feedback always welcome!
Similar to predictive lead and opportunity scoring, PAS serves to identify the best allocation of sales resources by collecting and analyzing thousands of data signals. The scoring process usually breaks down into three steps:
1. Ideal customer selection
The buyer identifies a list of "high-value accounts". This typically includes customers with a high LTV or a subset therein. The larger the sample size the better, so buyers often include high-value opportunities as well.
With list in hand, the vendor appends each high-value account with both internal and external data signals. This includes historical data from a CRM or Marketing Automation platform, in addition to firmographic and technographic data points mined from the web. Doing this creates what we call an ideal customer profile (ICP) -- or a framework for identifying new and existing accounts that are a unique fit for your business.
Featured resource: Learn how KnowledgeTree created their ideal customer profile.
3. Account selection and scoring
Using the ICP framework, a predictive model is built to identify accounts with similar makeup. Grade and/or number scores will be given to reflect how well each new account matches the ICP. Account scoring models are used to score both existing accounts and identify net-new prospects.
PAS is not for everyone. It relies heavily on the sophistication of your sales model and the maturity of the organization. Here's a logical progression of account selection tactics based on maturity level.
In building Datanyze's predictive solution, we took a slightly different approach than the rest of the market. Before setting out, we made two key assumptions:
1. Technology companies will be the primary users
According to a 2014 SiriusDecisions survey, 78% of predictive scoring adopters were classified as technology companies. This in mind, we decided to give considerably more weight to the technographic inputs (i.e. technology stack data) over other, more commonly used firmographic and historical data points.
For tech companies, technographics often provide a deeper, more sophisticated look into an account than do other company data points like revenue, industry and size.
2. Internal CRM/Marketing Automation data can be misleading
After talking to our customers, one thing became abundantly clear -- sales reps hate entering data! This makes capturing accurate historical information from a CRM or marketing automation platform quite difficult for any predictive solution provider. To combat this, we chose to focus on the data sets we could control, so as to eliminate any misleading inputs that could negatively impact the model. This means focusing more on external data points that we could find and verify by crawling the web on a daily basis.
Thinking about trying a predictive scoring solution? Here are a few common use cases taken straight from our customer base.
Supporting new product launches
About to release a product that will significantly increase your addressable market? For companies looking to develop targeted outbound processes around a new ideal customer, a PAS solution may be your best bet for quickly identifying the best accounts. Datanyze has no limits on how many models you can create, so customers can build models each time a new product comes to market.
Uncovering "white space"
No matter how good your sales team is at sourcing new accounts, there will always be many stones left unturned. By matching your ideal customer profile across our database of 45 million websites, useres can quickly identify gaps in their existing target market and make sure the best prospects receive consistent sales attention.
Identifying your "ideal customer tech stack"
A lot of PAS solutions choose to stay within what is called a "black box". This means they prefer not to share the positive or negative attributes of a model, because it's part of a secret sauce methodology.
On the contrary, we believe it's extremely important that customers get to see the inputs of every model, as it often helps reveal certain characteristics about ideal accounts. Since our models are built primarily using technographic inputs, the model summary helps our users identify which technologies their best customers are using in tandem with their own.
Feel free to get in touch with your CSM or send us a live chat today! We look forward to hearing from you. Need additional resources, check out our Co-Founder's recent presentation, Finding Your Best Accounts in the Age of Data and Predictive Analytics.
About the Author
Sam is the director of marketing at Datanyze. He's a big John Hughes fan who occasionally fills the DZ office with the sweet sweet sounds of 90s rock giant, Creed.Follow on Twitter More Content by Sam Laber