Google Analytics Cross Channel Attribution Models Defined
Gone are the days where statements like “Half the money I spend on advertising is wasted: the trouble is I don’t know which half” by John Wanamaker still hold water.
Digital advertising compared to traditional marketing models has opened up measurement protocols where we are able to measure almost “everything” from consumer behavior and affiliate this with brand advertisement to create a clear picture of ROI in terms of advertisement and brand communication. Digital marketing relies heavily on consumer data/insights and real time performance to determine the success of campaigns which are run across various channels with a set of measurable objectives and KPIs. So how do we measure the channel mix in a campaign and allocate subsequent budgets equitably to ensure greater returns and lower CPAs? Google Attribution Model via Google Analytics comes in handy.
According to Convertro, “Attribution is the science of assigning credit or allocating dollars from a sale to the marketing touchpoints that a customer was exposed to prior to their purchase”. In digital media context, I would define attribution as a set of rules that will determine which channels receive credit for conversions (micro & or macro) whether goals, sales or events that we have identified as having a desirable and measurable impact to our campaigns.
There are a number of Google Analytics Attribution models that one can deploy as measurement protocol to a campaign. To effectively choose one attribution model against another as a means of measuring the performance of various channels, a clear understanding of the business objectives as well as marketing goals is paramount. Below are examples of Google Analytics attribution models that one can use. Honestly, I personally do not vouch for any model as I go on record as one person who chooses one model, justifies it only to ask a myriad of questions once the data sets in. (Will highlight one scenario at the end of this article).
The Last Interaction Attribution Model: This model puts all emphasis on the last interaction that led to the conversion thus the last touchpoint is awarded 100% attribution. In the above example, the direct channel will be awarded 100% credit for the conversion.
The Last Adwords Click Attribution Model: Here the paid ad touchpoint receives all the credit. This model ignores all the other channels and is more valuable if you want to measure your actual ad performance in terms of CPA thus in the above example the search ad gets all the credit. Note that the last search ad gets all the credit assuming that two different queries triggered two different ads within the campaign.
The First Interaction Attribution Model: The first interaction within the conversion path receives 100% credit in terms of attribution. In this instance, the Facebook promoted post gets all the credit for the conversion.
The Linear Interaction Attribution Model: In this model, all the touchpoints within the conversion path share credit for the conversion equally. If there were four touchpoints within the conversion path, each channel receives 25% credit for the conversion as in the instance stipulated above.
The Last Non Direct Click Attribution Model: In this model, direct traffic is totally ignored and the conversion is attributed to the last touchpoint interaction before the actual conversion. In the above scenario, the remarketing ad will receive 100% credit for the conversion.
The Time Decay Attribution Model: In this model, credit is awarded based on the recency of interaction towards conversion. More weight is awarded to the last interaction based on time. Time lapse is inversely proportional to the weightage awarded per touchpoint thus in the above example direct channel will get the most credit with the Search ad & Facebook getting the least for the conversion.
The Position Based attribution Model: In this model of attribution, there is a fixed value that is assigned to both the first and the last touchpoints with the other touchpoints in between the conversion path receiving an equal distribution of the remainder. Usually the fast and last interactions are awarded 40% each attributed to the conversion with the remaining 20% distributed equally between the middle touchpoints. This will mean that Facebook and direct channels will receive 40% credits each with the search and the remarketing ads receiving 10% each respectively.
Finally, you can go further to implement attribution modelling cross device, offline to online and combine with the cross channel attribution above and you end up with this complex web of data that sometimes makes sense usually not... you wish you never told the client about all this in the first place. Bottom line is hopefully you may end up with a statement, "at least 75% of the budget was effectively utilized and one quarter was not. Good thing is I know which quarter".
Assume the Multi-Channel Funnel reports (MCF) (which we will look at in the next post) reveals that direct channel had the highest percentage in terms of top conversions per channel report followed closely by the remarketing display ad. Facebook as a channel has the best assisted conversion compared to all the other channels and the CPA for search was the lowest compared to rest. I begin asking myself such questions as;
- Why should I use last click attribution yet I was running paid media across all the other 3 channels except direct?
- Should I allocate more budget for the remarketing and slightly cap Facebook budgets yet Facebook assists most of the conversions and builds my remarketing list?
- Should I use linear attribution and treat each touchpoint equally considering I feel cheated by direct since conversions only increased since I started running the ads on the other channels?