1. Help Center
  2. Using boodleAI's Software

Common Prediction Issues and Solutions

You have problems? We have solutions.


Some organizations encounter issues on the way to achieving the full benefit of boodleAI’s predictions. While this occurs sometimes, there are proven solutions to the issues you may run into. These solutions will help achieve the lifts in acquisition, engagement, and retention you desire.

Successful predictions require the following: (1) selection of the predictions appropriate to the campaign; (2) selection of guidons that yield the appropriate predictions; (3) selection of training data that builds appropriate guidons (and iteration of the guidon if the initial training data proves insufficient); (4) adequate sample size to provide measurable results; (5) selection of quality target data; and (6) a fundraising campaign that produces measurable results with the targeted audience.  If any of these elements are missing, then your predictions may not unlock the full potential of insights hidden in your data.

If you encounter one of the below issues, please contact the boodleAI Customer Success team (success@boodle.ai).  We’ll work closely with your team to develop and implement potential solutions. 

Prediction Issue


Potential Solution

Prediction Selection

Where there are multiple guidons used to filter a prospect list, a Prediction Selection issue arises when the guidons are appropriately selected and trained, but a different set of predictions should be used for the campaign.

Review the predictions and choose a different set of predictions by using a different combination of guidon scores to filter the prediction.

Guidon Selection

A Guidon Selection issue arises when the guidon(s) is/are appropriately trained, but different guidon(s) should be used for the campaign

Review the available guidons and select a different combination of guidons to produce the predictions.  Some possibilities: (1) use instant guidons instead of custom guidons;  (2) use a different custom guidon; (3) combine custom and instant guidons.

Guidon Training

A Guidon Training issues arises when the appropriate guidon(s) is/are selected, but different training data should be used to train the guidon(s).

Review the training data and (1) select a different positive training data set; and/or (2) select a different negative training data set. 

Guidon Iteration Required

Guidon Iteration is required when the appropriate guidon(s) is/are selected, but another iteration of the guidon is required.

Review the results from the campaign and supplement the positive or negative training data to improve the existing guidon.

Sample Size Issue*

A Sample Size Issue arises when the number of predictions used is too small to provide a meaningful test of the guidon(s) given the assumed response rate.  In other words, more predictions are required to allow for a meaningful response

Select a larger sample size.  Please see this article to determine the appropriate sample size. 

Target Data Issue*

A Target Data Issue arises when the target data set consists of leads not likely to respond to the campaign at the desired response rate.  In other words, different leads are required to respond to the selected campaign. 

Select a different target data set (prospect list). If response rates remain lower than expected, there may be a Campaign Performance or Sample Size Issue. 

Campaign Performance Issue*

A Campaign Performance issue arises when the campaign selected is not likely to produce the desired response in the selected leads.  In other words, a different campaign is required to produce the desired response from the selected leads.

Select a different campaign.  If response rates remain lower than expected, there may be a Target Data or Sample Size issue.

* Note that use of a campaign with a known baseline or use of A/B testing can eliminate Sample Size, Target Data, and Campaign Performance as potential issues.

If you're ready, here's How to Use Predictions. If you have any questions, don't hesitate to reach out to us at success@boodle.ai.