Frequently Asked Questions
1. What is a Custom Event Model?
In Atlas, a propensity model built by uploading a list of Accounts with data about whether they experienced a significant event on their journey with you is called a Custom Event Model.
Common uses for this kind of model include:
- Account Fit models
- Product Fit models
2. What if my target segment has both prospects and customers in it?
All Accounts in your Segment will get scored by your Custom Event model, so be careful that you run the right campaigns against your Accounts to connect with them where they are in their journey with you.
Best practice is to separate these into separate Segments so that you can easily use the scores from your Custom Event Models. Do this by creating separate Segments for prospective new business versus relationship growth bases, etc.
3. What if my target segment doesn’t have enough Accounts to train a model?
Lattice recommends using at least 10,000 Accounts when training a custom event model. With clean data and high match rates, this can be pushed down a bit.
Still, sometimes more data is needed. If you have not already done so, try enhancing your segment accounts with data from your lead database. Atlas make this easier by enabling you to filter out personal email domains if they are in your file but don’t have company information.
When you simply can’t find enough data within your Segment, try expanding around your Segment definition enough to build a model.
4. Can I use my transaction data in Atlas to create my training file?
When your Spend data is available in Atlas, you can use Has Purchased attributes to create lists of Accounts that have been targeted successfully. You can export these lists to build your training file.
5. How does one record per account work when I have more than one value for Event?
One record per Account will always try to choose a successful event for your Account when one is available.
6. How can I use opportunities to create a good Event column?
Pull a list of Accounts with Open Opportunities. Include opportunity Stage in the pull. Now you can experiment to find the stage that will give you a good picture of top versus bottom funnel to build a Fit model.