What is a predictive scoring model?
A predictive scoring model uses machine learning technology to score accounts based on a “training” dataset of historical information. It looks for patterns in the training dataset to create the model and then applies that model towards another dataset that needs to be scored. For RollWorks Account Scoring, the training dataset is a list of customers and the model is applied to a score a target account list.
What information do I need to provide for the training dataset? And what represents a good training data set?
The predictive model will be only as reliable as the training data used, so we recommend providing a list of 200 customers that you consider a good customer set as the training dataset - we only need domain names.
If you do not have 200 current customers, you can supplement the training data set with accounts in the sales pipeline that have been qualified as good accounts by your sales team. For example, accounts with open opportunities against them or accounts in the proposal stage.
Is your scoring model based on fit, third-party intent, or onsite engagement?
The RollWorks account scoring model is based on static account fit data - firmographic, technographic, geographic and social data. At this time we do not include any intent or engagement data in our scoring model.
What does it mean to be a more highly graded account?
After a list is scored, RollWorks will return a grade range from A to F. A higher graded account shares more characteristics (or fit attributes) with a “good” customer profile built by the predictive model than an account with a lower grade.
Can I push the account grades into my CRM?
Yes, account grades can be exported to a customers’ CRM (only Salesforce).