Improving our Marketing / Lander Algorithm using Prior Sales Data

Project Description

We are an online entertainment and media company that sells memberships to our content collection online. (i.e. Netflix)

Our goal is to increase sales by improving the algorithm that determines the order of thumbs that our lander shows.  Additional goal is to come up with a automated system to introduce new thumbs to the lander.

Our lander: http://cdn.teamskeetimages.com/template_tour.jpg 

Our join page: http://cdn.teamskeetimages.com/template_join.jpg

Sales Funnel:  

A. Click on advertisement links / type in

B. Tour page / click on thumbs

C. Click on join page (Fill out join page)

D. Biller page (fill out and submit)

E. Approval Page

Data we collect:

1.  Traffic / impressions

2.  Sales 

3.  CTR (Click Thru Ratio)

4.  Time 

5.  Spot (location of thumb)

6.  Type of device (desktop or mobile)

Our current tour algorithm is:

75% x Average daily joins + (25% x CTR x added factor)

Where added factor is a number that we increment 0.02 every 3 thumbs.

Example

1.00 for thumbs 1 to 3

1.02 for thumbs 4 to 6

1.04 for thumbs 7 to 9

Misc Points to Consider:

A.  Our surfers come from all across the world (English speaking countries is our focus) and thru all different sites (including Tube sites, Review sites, etc) - The quality of varies greatly from region to region to type of sites the surfers are coming from.

B.  Added Factor is used in our algorithm with the thought that the lower the thumb on the list - the less chance it has of being seen 

C.  Custom thumbs - one of the most important points.  We add new thumbs on a daily basis to the system.  On the above image, we use Thumb 5 / 10 / 15 as custom thumbs, we let these spots get traffic for a few days to give it a push, then release it to the tour and let the thumb end up where it may. One of the secondary goals is to have an optimal, automated strategy to test these custom thumbs.

D.  Our landers can vary how many thumbs are shown depending on how big the consumer monitor is - lowest being 2 - highest being 6.