Google Releases Data-Driven Attribution Model
- by Laurie Sullivan @lauriesullivan, August 20, 2013
Attribution modeling tells a great story on the consumer's path to conversion, but the biggest problem for online advertising has been the missing technology and the data to make it a reality.
Google made a major shift toward attribution marketing Tuesday with the release of a premium analytics model that relies on economics and statistical analyses to analyze data across search,
social, display, email and other media.
Available globally to all Google Analytics Premium customers, the tools analyze the entire customer journey, whether that journey ends in a purchase or not. It automatically assigns values to each media to give marketers a more complete view of the digital channels and keywords that are performing best and compares non-conversions to provide a more accurate comparison. The model analyzes "similar" site visitors based on a variety of channels, ad exposures, ad placements, ad types and repetition.
Statistical analyses and economics modeling predict the actual behavior of consumers. Clients can use the platform to select and analyze any online media. Attribution reports allow marketers to compare values, as well as sort and filter data to identify changes that impact campaigns. The tool also integrates with DoubleClick Digital Marketing, the Google Display Network, YouTube, and virtually any digital channel.
Google's platform compares and determines the likelihood of conversions given a specific order of events. The difference and probability in path structure becomes the foundation for the algorithm, determining the weight of each channel.
That has been the biggest challenge: determining how much one channel works toward contributing to the purchase or conversion. During the past two years, Google worked to build a foundation in multichannel funnels and attribution comparison tools. All that comes together with algorithmic models and a new set of reports designed to take the guesswork out of attribution.
While many Google tools are free, this one comes with a price. Google Analytics Premium charges an annual fee for accounts with 1 billion or less hits per month and tiered pricing for more.
Interesting from a technological development standpoint. However, getting attribution measurement (this already has black-box tendencies) and Web analytics from the same company that sells you display media and search is a simple conflict of interest...no mention of firewalls or data protection and everybody knows that Google Analytics cookies is the same one used by Invite Media/Bid Manager.
Oh well, a Fool and their data are soon parted: http://goo.gl/pd5ACa
Thank you Laurie,
Your article is very educational and very easy to read. As you mentioned at the end of your article, Premium is not cheap, but I think it’s worth each penny. The way I see it, to create a good attribution model, company needs to hire a statistician/modeler (in some cases more than one), it can be very expensive and let’s not forget that it takes time to find statistician/s, he/she needs to familiarize himself/herself with company’s data and only then build the model (and it might not be perfect after first time). Also a positive thing is that Premium allows company to dig into data driven attribution model if it really wants to understand why and how credits were distributed between touch points. I think it’s a big plus.
Thank you again!