A Simple Way To Think About Attribution

Sharvanboskirk [Posted by Shar VanBoskirk]

My colleague Emily Riley** has written about attribution modeling -- the "new" approach to online measurement which tracks more than just the last ad clicked.  Emily's approach of "block and tackle attribution" gives marketers a framework for how to track value across both mainstay and social interactive tools.

Apart from Emily's research though, I haven't really heard a lot of structure around how attribution works.  It seems like one of those things that everyone is buzzing about, but few have any real discipline around. Well, yesterday I sat in on a panel on attribution which simplified it enormously.  

Madan Bharadwaj, the architect of Bluestreak's data analytics solutions boiled attribution down to three steps:

1) Organizing user path data -- determining the path a user takes to purchase
2) Contribution modeling -- evalutating the different influences along this path
3) Cross-channel optimization -- applying these insights to make media allocation decisions

I know that the steps and technology to do these three things are complex, but Bharadwaj encouraged audiences to think simply in order to reduce unncessary complications.  His advice: "Start with one question that you'd like to get the answer to.  This will help you filter through the volumes of data you can collect to instead focus on the data that matters to that question."


Adam Goldberg, the chief innovation officer for advertising measurement platform ClearSaleing also encouraged marketers to start attribution even if only in a small way, and to find value in all insights while still working to refine them.  In his words, ""Don't let perfect get in the way of good."


**Emily will be releasing a Wave on attribution modelers, which includes ClearSaleing, shortly.


re: A Simple Way To Think About Attribution

This is the same method that Atlas has been offering as part of their ad serving suite for over 2 years. Not sure what the industry push back on adoption has been.Certainly other marketing touches have an effect on a conversion at a search engine,(or wherever the last touch occurred.)

re: A Simple Way To Think About Attribution

Totally agree that the notion of attribution is not new. I just sat in on a case study presentation by Reliant Energy that showcased how they have been doing attribution offline for years modeling the effectiveness of their offline media, telemarketing and direct mail at driving the eventual close. I think the recent fascination with it online is due to the huge investment marketers are making in search marketing, and also the emphasis on accountability of all forms of online media. Traditional banner ads may not be the best at driving the end sale, but I think marketers are finally looking for metrics to defend that they do *something* that influences an eventual conversion...even if that conversion occurs in through another medium.

re: A Simple Way To Think About Attribution

With Internet marketing becoming more holistic, online attribution (aka market mix) makes total sense. I believe the interest has been escalated given the economic downturn affecting marketing budgets requiring total accountability - or close to it. I'm sure various models will continue to evolve as digital marketers will realize it's importance. But as you've mentioned, in the interim it would be advantageous to implement attribution in simpler terms first... in other words, walk before running!

re: A Simple Way To Think About Attribution

I think you are right, Dan. People have to start somewhere and sometimes the complexity of everything people could or should be measuring is paralyzing.

re: A Simple Way To Think About Attribution

More than happy to show where we are helping to implement many of these systems through our 'universal tag'. Because every tag 'runs' through our system - we can provide that full user journey view.

re: A Simple Way To Think About Attribution

I agree the concept of attribution is not new but unfortunately there are still many issues that need to be addressed, such as...1. Ad servers reliance on their tag to be served on the last visit preceding an action. Unless I'm mistaken, ad servers above can only attribute credit for prior engagements if the last click preceding the conversion is goes through their server. Too bad most conversions are directly preceded by visits from direct nav and/or natural search.2. Agreed upon methodology for recasting the cost per action across the touchpoints that played a supporting role. How far back do you go? How many impressions are worth one click? How do you split the credit across different types of media?3. Widespread acknowledgment that our ability to measure impact is severely limited by increasing use of multiple devices (work, home, mobile) and cookie deletion. For every action we can measure via cookies, there must be 3-4 that we can't measure.I'm happy to hear about solutions arguments to the contrary! BTW feel free to critique my views on measuring impact of online demand generation at http://bit.ly/j5xtf.Steve Lathamhttp://twitter.com/stevelatham