Overlap amongst Acquisition Sources

gaurav2

18-02-2010

One of the most important insight that all CMOs (and thus analysts) need is the RoI of each of the acquisition sources. With multiple campaigns and SEO running simultaneously, true understanding of RoI is dependent on understanding the overlap amongst these sources. The first-click, last click and linear success event attributions are far too elementary to bring out any insight about overlaps. Also, Unified Sources DB Vista and cross-visitation (campaign stacking) plugin have significant limitations (it is their classifications that the pathing/overlap is needed on and not the tracking codes themselves) leaving a lot to be desired.

 

I have a couple of visualizations that I can imagine:

 

1. Venn diagam on number of visitors showing intersections of different sources that we select.

 

2. Pivot table view with both axes showing acquisition sources and common visitors shown on each intersections

 

3. Pathing or hierarchy view where both visitors and success events are shown at each step. For example - Paid Search (10000 visitors, 100 orders) >>> Natural Search (2000 visitors, 200 orders) >>> Direct/none (1000 visitors, 300 orders) would mean that 10000 visitors that were acquired from Paid Search first time, gave 100 orders, but 2000 of these visitors came to the site using Natural search in some subsequent visit and gave 200 orders. Out of these 2000, a 1000 came directly in further subsequent visits and gave 300 orders).

 

I think this topic required a 3 page blog post, which I hope I would do some time soon. Till then (particularly in case I have lacked clarity in the idea) please feel free to reach me at

gaurav.iitk@gmail.com or gyanalyst@twitter

2 Comments (2 New)
2 Comments

adamgreco0

21-02-2010

I think there are many ideas posted here about multi-session campaign attribution so I am sure Omniture will get the idea and combine them as necessary.  I have advocated a way that SiteCatalyst users could choose how they want to assign attribution (i.e. 50% 1st touch, 25% Last Touch and divide up the rest evenly...).  It would be great if there were that kind of flexibility since in today's landscape it is really hard to determine which value should get credit.  I also don't think this is limited to just Campaign elements...

gaurav2

21-02-2010

Agree with you on the suggested success events attribution. However, that still does not solve the problem of overlap amongst sources in terms of visitors - even when they have not reached any success event - Just to understand the extent of 'redundancy' amongst campaigns (same visitor acquired through multiple campaigns) and  also to understand some correlation between campaigns (what fraction of visitors who responded to iPod campaign also responded to iPad campaign).