After creating your prediction, you have three segments available on thalox.
Learn how to push your segments into HubSpot.
thalox gives you a lot of insights into your HubSpot contact database.
1) To see the fields that are contributing to your model, go to your predictions tab and select your latest prediction.
2) Here you will see a tab which is called “Statistics”.
It is very interesting to see those numbers over time. The main goal is that the number of contacts without e-mail activity is decreasing and the average engagement by contact is increasing.
But there is much more thalox can show you.
3) Once you are on the “Statistics tab”, please click on the button “Download table as CSV file” and open the CSV file with Excel or Numbers.
| Note: The file is using “Comma” as a separator. |
4) Once the CSV file is open, please add a filter and select the field “ml_inclusion” with the value “true” as shown in the example.
All those fields presented in the CSV file contribute to your model and will influence the Thalox Engagement Score (TES). on which the three segments are built.
Very often we get feedback that marketers never think about filtering those fields to create their lists.
| Note: Those fields could change when doing a new prediction since thalox is learning from the past and is always recalculation based on your previous campaigns. |
“Variance Percentage” field:
A very interesting field is the “Variance Percentage” field. It shows how many different field entries you have. The higher the number (Maximum 100) the more different the field entries you have. For an ID or E-mail field, 100 is a good value, but for Job Title, the “Variance Percentage” field should be very low so that you could filter on those fields.
We advise you to take a look into your HubSpot Instance to normalize those fields if needed.
“Null Percentage” field:
The field “Null Percentage” gives you the fill rate of a field. The maximum is 100, which means that the field has no entries in it. Review your field entries based on that information. In most cases, you will find out that normalizing your data could help you to get better results.
