Attribution and the art of understanding where customers come from
Stephan Brisard - 11/11/2018
For many years now we have been talking about attributing marketing success to the right channels. Attribution models have flourished and gotten more and more complex as the ecosystem of marketing touch points grew. But the question remains: which attribution model is the most successful to drive meaningful conversions which tie to greater return on marketing investment?
Today thanks to Google Analytics or Adobe Analytics, we can find out pretty much everything about anonymous, known and repeat visitors coming to a given website. We can track down all the various sources and assets that contribute to conversions (leads, customers, repeat buyers etc.). What we’re struggling with though is to understand which sources or channels have the highest impact on conversions? What behaviors and patterns also truly motivated a user to commit and action on a website?
There is a lot of confusion still between owned, earned and paid (social) media as to the weight each brought upon an actual conversion. I have been using additional tools like Oracle Eloqua or Marketo to try to figure out which marketing channels had the greatest influence on an actual conversion and garner additional data points that would allow to optimize my various programs. There are many factors to consider as part of the customers’ journey which include the content they viewed, the assets they downloaded, their signed requests for demos or their buying patterns. Just tracking the conversion itself without the historical background that precedes it does not make much sense.
It’s amazing to me how many organizations I have seen that just don’t seem to care or simply have never heard of an attribution model. Some will argue that it is part of testing and that driving revenue is all that matters; true, yes performance is attributed to revenue generation but understanding how customers have been generated is equally important to properly optimize the right channels and maximize marketing expenditures towards better outcomes.
It usually takes about 7 to 10 touch points really before you can start measuring an actual conversion; it means that customers have to interact that many times with a brand before making an online or offline purchase. Touch points can take the form of an email campaign, a social media campaign, a PPC ad or other marketing mechanisms that will guide intent and ultimately influence purchasing decisions. The question again is how do you properly attribute the actual final conversion to the right touch point(s)?
Today there are several attribution models available; some are more used than others and you’ll find that the most sophisticated marketers will leverage several models at the same time. Let’s take a look at the most popular ones.
- First click
First click attribution gives all the credit to the touch point that first drove a potential customer to the right content on a web property. It’s a good indication of where it all begins but does not take into consideration the entire journey of that customers; what happened before, in the middle and after?
- Last click
This is the model that I have seen the most used. In this scenario the attribution is linked to the last touch point that will drive a conversion. The problem and as it was mentioned above in the post is that most customers and prospects will usually interact multiple times with a product or brand before finally make the last leap towards a buying decision or an action e.g. requesting a product demo. It’s a short sighted mechanism to truly understand the value of the conversion and how to further model an optimization plan.
- Time decay
The core premise of the time decay is that the touch point closest to the conversion gets most of the credit, and the touch point prior to that will get less credit based on a smart and simple algorithm. This method tends to minimize the first touch point, but it does begin to address the importance of in between steps. It’s not a bad method, since the further back a touch point is (take SEO or Paid Social for example) the less credit it should get.
This model uses a weighted approach to split credit equally towards each touch point. While this model helps mitigating risk, it’s somewhat an arbitrary approach which does not really help with properly optimizing various touch points, marketing campaigns or programs.
By default, the Position Based model attributes 40% of the credit to the first and the last interaction and the remaining 20% is distributed evenly to all the interactions in the middle. I like this model because it’s simple and provides equal weight the first and last touch which makes it easier for brand recognition.
- Personalized attribution
It’s a cool and probably much better model than any of the others cited previously. Google Analytics allows today with its free version to develop your own logic. This model allows the use of the Linear, First, Last, Time Decay and Position Based models as a starting point, and then layer in other factors considered important for a business to create its own attribution model. In this model, one can choose to put more weight on a certain touch point based on the known propensity for a certain target audience to be more receptive to certain content types or marketing vehicle e.g. email, PPC, paid social, banner ads etc.
7. Predictive analytics attribution
This is the future and the holy grail of attribution. Very few companies or even technologies are capable to doing this well just yet. With the right tools, engineering minds and tons of set-ups one could gather enough data and set-up rules with a variety of tools that would suggest the most accurate attribution models. Predictive analytics models typically look for trends, patterns, competitive analytical sets, cross reference data points and end up determining the best conversion path for a company to adopt and that ultimately will optimize sales.
Hopefully this post has shed some light into what’s available today to properly track the effectiveness of various marketing channels and be able to properly optimize based on a better attribution of various marketing touch points. The first click and last click models are flawed and obsolete; modern marketers have to challenge themselves and thing of better way to leverage available tools on the market today and really invest the time to set-up more comprehensive models that will bring value to the right set of touch points.