Emails delivered on time to each customer

Problem:

Email open rate was lower than it should have been

Cause:

Lack of information on email delivery times and open rates

Solution:

Dynamic scenario timed to wait for a specific event, e.g. time of last purchase or time of last delivered push notification.

Deliverables:

Created a scenario which used the ideal email time algorithm. This started with the setting the ‘On date’ trigger and adding a Wait nod. The ‘ideal_email_time’ customer attribute was extracted with Jinja: {{ customer.ideal_email_time }}

Feature was A/B tested with a split before the custom Wait nod, and the Control Group was let to wait for a specific amount of time (that matches the regular sending pattern).

The model was split into two separate models, one with weekday data and the second one with weekend data as the Open Rate statistic performed better in the Control Group during weekend days.

The optimal time can differ hierarchically according to the following points:

1. Individual auditory
2. Day of the week
3. Time mailing was performed
4. Topic of the mailing