Have you noticed that a record's Customer Fit score or Likelihood to Buy score has changed and would you like to understand why? This article can help you understand common reasons for score changes.
The Dynamics of Customer Fit, Likelihood to Buy, and Lead Grade Score Changes
Generally, MadKudu's Customer Fit fields will rarely change over time unless there is a model update. They are based on static attributes such as technographics, demographics, and firmographics, in contrast to the MadKudu Likelihood to Buy and MadKudu Lead Grade fields which are likely to change every day as they are fully or partly based on behavioral activities. Even in the absence of activity, the Likelihood to Buy and Lead Grade scores will change because of decay applied to MadKudu behavioral scores.
Why Did My Customer Fit Score Change?
You may still notice changes in the Customer Fit of a lead or an account on a daily basis due to two different scoring methods used by MadKudu: real-time and batch scoring.
- Real-time scoring, available only for Customer Fit, updates the score within 5-15 minutes of the lead's creation in your CRM. Real-time scoring is based on the standard enrichment MadKudu has on this email from their own sources (e.g. Clearbit, HGdata, PredictLeads).
- Batch scoring updates the score every 4-12 hours with any new information. Batch scoring updates the score with additional enrichment available from your CRM about this email or the domain the person is attached to. Therefore, the score of a lead can change due to new information or activity being added to their record in your CRM.
Best Practices to Limit Score Change in Customer Fit
Note that if your Customer Fit model uses CRM fields to score leads and accounts, MadKudu may not receive the complete scope of information in real-time. In real-time, the data points might not exist yet when scoring records. When MadKudu receives the complete scope of information in batch from your CRM, it updates the score.
When using CRM enrichment in your trees and overrides, pay attention to the fields that the model will use to score leads in real-time. Always make sure that the real-time score is lower than the batch score to avoid downgrading a lead's score in batch, which could potentially remove them from the list of MQL leads for your sales team. This could indeed raise questions from your sales team.
Here are a few other best practices to keep in mind:
- Avoid using split conditions based solely on CRM enrichment fields in the top nodes of your trees. This will help ensure that enrichment doesn't cause a big score change in batch.
- If you're using combined computations, make sure to leverage MadKudu enrichment before your own CRM enrichment.
How to Use the Score Lookup to Understand the Difference Between Batch and Real-Time Score
The Score Lookup page makes it possible to understand the score a lead will get in real-time, based on the enrichment immediately available for scoring in MadKudu. If you use CRM fields in your model, this score can be different from the score in your CRM (which is the batch score).
Still have questions? Feel free to open a ticket here.