The modeling process - an overview

The modeling process

1. Dataro integrates with your system and establishes regular extracts of required data
   1. Complete process: Dataro captures all records for the required parts of your database. This is done initially during onboarding and then at regular intervals (typically every 4 weeks). This is necessary to capture deleted records, which are not reflected in the delta process.
   2. Delta process: Dataro captures the “delta” of data which has changed since the last update. Typically this is performed each night. The delta extract is merged with the previous complete snapshot to produce an updated and complete replica of your data.
2. Your data is cleaned and normalised to the Dataro Reference Schema, the common data model that we use across all organizations. 
3. During onboarding you will be required to complete the campaign tagging process. This is important so we can understand the category and channel of each gift and communication in the system.
4. We transform your data into a set of hundreds of different numeric “features” suitable for machine learning. Learn more about our model features here.
5. We submit the featurised datasets to our predictive models to produce predictions. The predictions give us propensity ranks and scores for each donor in your database, showing you how likely individuals will be to take certain actions and how they compare against the rest of your donors. Read more about ranks and scores here.
6. Predictive Ranks and Scores are published to the Dataro Platform and, in some cases, your CRM (depending upon your connection type + package).
7. Dataro routinely reanalyzes your data and updates the ranks and scores as you continue engaging with donors and generating new fundraising campaigns. This is typically done on a weekly basis.