Multi-Click Attribution: Tracking the Way Conversions Actually Happen

By Daniel Riveong

Interactive marketing has long sold itself on the promise of accountability and ROI measurement. Yet, at the same time there continues to be challenges in solving John Wanamaker’s problem: “Half the money I spend on advertising is wasted; the trouble is I don’t know which half.”

Indeed, the common analytics tracking for online marketing campaigns are simplistic at best. It cannot measure how display ads (seen as branding) can impact search (used mostly for sales). Nor can most analytics tracking attribute the multiple searches and visits a person did that ultimately led to a purchase (See image above).

Enter Havas Digital and Yahoo. Their panel, Search + Display—Moving Beyond the Brand vs. Direct Response Model, not only showcased Havas Digital’s solution to this question but show great case studies of the insight and optimizations gained from a better and more true multi-channel analytics. (Note: all of the images in this post was taken by the Havas Digita/Yahoo PowerPoint presented at ad:tech)

MODERATOR:
Rich LeFurgy, General Partner, Archer Advisors

PANELISTS:
Dan Boberg, VP, Advertiser & Agency Professional Services, Yahoo!
Ed Montes, Executive VP, Managing Director US, Havas Digital US
The Promise for Advance Attribution in Analytics
The benefit of building a system that can do sophisticated attribution tracking lies in being able to track what are known marketing truths but have been difficult to readily measure:

1. Display (Brand) Supports Search (Sales)
There has been countless studies revealing that running display banners, which may or may not have been clicked on, increasing searches for the brand and eventually in to sales. As far as data go, the panelists quote two of the most often cited statistics on the matter:

1. Display + Search Provides Improved Branding
   “Exposure to a display advertisement increased related trademark term searches (brand, company or product names) by an average of 26 percent” – Yahoo/ComScore study of Fortune 100 advertisers, “Close the Loop: Understanding Search and Display Synergy”
2. Display + Search Provides Higher ROI
   “Users exposed to both search and display ads convert at a higher rate: 22% better than search alone, and 400% better than display only.” – The Atlas Institute, “The combined impact of Search and Display advertising – Why advertisers should measure across channels”

Having an attribution system that can readily attribute how much a display campaign drove search ROI would shift our perception for display as “just” a branding tool.

2. Tracking Beyond the Final Click
By default, most analytics systems attribute sale to the last link the customer clicks on. It completely ignores the fact that people may conduct multiple searches and clicks to a ecommerce website before than actually make a purchase.

All marketers know that the “last click” is not correct. If a search marketer blindly believed in the “last click” and on strict ROI measurement, we would all just bid on our branded terms and go home.

Case Studies
Working with Yahoo, Havas Digital did build such an attribution system that would help track the two issues above. Havas Digital and Yahoo presented several case studies on the system, called Armetis. Here are slides from a few of their case studies.

Case Study #1: When display drives sales-based ROI…for other campaigns

The above is a case study from an automobile manufacturer. While sponsored search drove the most “last click” conversion, the entire marketing picture shows that display advertising assisted in the conversion of 163 visitors. So scaling back down display advertising dollars may, in fact, cause lower sales for other marketing channels like search.

Case Study #2: Why some highly converting keywords cannot stand alone

From the above report, Campaign 2 seems to be both costly at a cost per conversion of $422 and one lowest leads performer. However by looking at search assist, we see it has been drive in assisting in leads for other campaigns. Without such data, a search marketer may have been too eager to kill campaign #2, yet end up driving lower leads for other search campaigns.
Closing Thoghts
While Havas Digital’s Artemis attribution system looks very impressive, there is no such thing as a perfect tracking solution. As the panelists pointed out, there are still large limitations based on the fact that users delete cookies and that cookies need to last (and survive) more than 90 days in order for tracking systems to properly attribute the multiple exposures and clicks that lead to a conversion. Then of course, comes the more complicated questions of how to weigh how far back can a banner show to user “claim” to have assisted in driving a sales one, two or five months later. The joke the panelists mentioned that one flaw is an agency could blanket the entire web with display banners and then claim “attribution” for any sale that happens there after.

In end, however, we do need tools like these to really optimize campaign and paint an ever more accurate picture of what really drives engagement and sales. Yet even the tool described here by both Havas and Yahoo were noted by both companies as still rough and far from complete. Indeed, the quest for ever better analysis is an interesting problem to me; I plan to write a follow-up to this soon on my own blog at Emergence-Media.com.