Survival of the "Fittest"
The predicted fit score can be best thought of as how much does a given record resemble records used to create the guidon (in other words, how much of a "fit" are they for an organization based on similarity to past and current individuals who are a fit for the organization). The higher the score, the higher the confidence boodleAI has that the record is similar to past and current individuals who are a fit. Those records with higher fit scores are more likely to be a fit because they are predicted to have a greater similarity to past/current individuals who are a fit and thus are worth more effort.
Note: the scores themselves are not a prediction of likelihood. In other words, a records with a 90 score isn't 90% likely.
Custom Scores are generated by Custom Guidons created from the organization's own first party data combined with third party data.
Instant Scores are generated by Instant Guidons created from third party data.
Note that all of the above models (custom and instant) maximize precision at the expense of recall. So, Guidon minimizes false positives at the risk of having greater false negatives. Therefore, while you can trust a high score (the precision is good), a low score doesn't necessarily mean the prospect is a bad prospect -- it may also mean the model couldn't score that prospect higher because of lack of data or fidelity in the features selected by the model as applied to that prospect.
Bottom line: when Guidon predicts that a prospect is good, you can trust that. It's worth noting that there are also good prospects in the lower scored records (but they are buried in a lot of true negatives).
If you have additional questions about predicted fit scores, please contact us at email@example.com.