How to Build a Predict Model

Now that you have been enabled by your account administrator* to create models, it's time to get our hands dirty and build! 

1. Click on the Predict tab

2. Select "Create New Model" 

3. Select the data import method:

Option A: Upload CSV 

1. Enter model name: 

 2. Upload a .CSV file with a list of domains or email addresses: 

3. Upload a .CSV with domains or email addresses of prospects which will be scored based on the model (optional).

B. Use SFDC data: In order to use this option a SFDC integration must be set up as shown below*:

*If an integration is not set up, you will be asked to set up a SFDC connection:

  1. Enter model name.
  2. Import the domains or email addresses of your ideal customers from SFDC:
  3. Import the domains or email addresses  of prospects which will be scored based on the model (same interface as above) (optional):
  4. When the results are ready, an email will be sent: 
  5. Model Results will have two tabs when the user has not skipped the third step (Upload Prospects) in model building process. If the user skipped the third step then only New Leads are shown:

  • Existing Leads: The scored and graded list of Prospects synced by the user in the third step of model building. This is helpful for users to know score leads which they are currently working on.
  • New Leads: List of leads from the Datanyze database which are good matches to your current customers

6. Summary Tab shows Positive and Negative attributes that were used to score this model. This includes both firmographic and tech data. The attributes will be different for each model:

Best Practices:

  1. Logical Segmentation- Size or Industry or Region
  2. Find 100 or more good existing customers for each segmentation
  3. Review model for each segmentation- Use Summary tab for positive/negative attributes.
  4. Use the Thumbs up or down to refine the results.


*Your account administrator will need to enable your account to create models. 

Was this article helpful?
1 out of 1 found this helpful
Have more questions? Submit a request