Expand my Community achievements bar.

Applications for the 2024 Adobe Target Community Mentorship Program are open! Click to the right to learn more about participating as either an Aspirant, to professionally level up with a new Certification, or as a Mentor, to share your Adobe Target expertise and inspire through your leadership! Submit your application today.
SOLVED

Adobe Auto-Target Result Explanation

Avatar

Level 1

Hi Team,

We have build Auto-Target activities and some have had positive lift and some has negative lift. All activities had the green check icon next to each experience. However, in our report, we haven't reach 95% confidence level. My questions are as follow:

1. For an activity with positive lift for all experiences, should we wait for until 95% confidence level to say the model works, or don't need to wait and we can say the model already works. We can split more traffic into experiences than control? (The confidence level always has fluctuations.)

2. If In a Auto-Target activity, one experience has positive lift, the other one has negative lift and all with green check icon there. How could we explain that? Could you please reply with two situations as:  reach 95% confidence level and below 95% confidence level;

In short, what is the best practice to explain the confidence level and lift in auto-target activity.

Thanks and Regards,

Jack

Topics

Topics help categorize Community content and increase your ability to discover relevant content.

1 Accepted Solution

Avatar

Correct answer by
Administrator

[3.30.22 Adobe Target Community Q&A Coffee Break Answer provided by @ryan_pizzuto:

----------------------------

 

hey @jackz11447106 great to see you are in the wide world of machine learning and thanks for the question in the Community!

When it comes to Auto-Target and Automated Personalization, the green check marks that appear next to the experiences/offers are not an indication of lift & confidence...they are simply and indication that Adobe Sensei has successfully built a model for those experiences. As we know, the more information a model gets, the better it becomes. So think of the appearance of the green check mark as the beginning of the reporting period. Before those check marks appeared, all experiences were served at random to get Sensei the data it needed to start optimizing. 

Pro-tip: make note of when models are built for all experiences, then change the start date of the reporting to when you had models built for all experiences. Note the difference in lift/confidence. Then KEEP GOING! The models get better with time. So think of the green check mark as the beginning of the process, not the end. 

----------------------------

Check out the entire 3.30.22 AT Community Q&A Coffee Break discussion thread for more insights on related topics

View solution in original post

1 Reply

Avatar

Correct answer by
Administrator

[3.30.22 Adobe Target Community Q&A Coffee Break Answer provided by @ryan_pizzuto:

----------------------------

 

hey @jackz11447106 great to see you are in the wide world of machine learning and thanks for the question in the Community!

When it comes to Auto-Target and Automated Personalization, the green check marks that appear next to the experiences/offers are not an indication of lift & confidence...they are simply and indication that Adobe Sensei has successfully built a model for those experiences. As we know, the more information a model gets, the better it becomes. So think of the appearance of the green check mark as the beginning of the reporting period. Before those check marks appeared, all experiences were served at random to get Sensei the data it needed to start optimizing. 

Pro-tip: make note of when models are built for all experiences, then change the start date of the reporting to when you had models built for all experiences. Note the difference in lift/confidence. Then KEEP GOING! The models get better with time. So think of the green check mark as the beginning of the process, not the end. 

----------------------------

Check out the entire 3.30.22 AT Community Q&A Coffee Break discussion thread for more insights on related topics