Wednesday, February 13, 2013

Deciding Who Wins

Attribution Modeling

        Attribution Modeling is not necessarily a fascinating topic, but it is very useful to know.  I would compare it to making pancakes; it's not exciting or very difficult to understand, but everyone should know how to make them.  You may be thinking, "what in the world is attribution modeling?"  I had no idea what it was before researching for this post.  Basically, it is the use of different models containing a set of rules that are used to determine who gets credit for conversions.  This can be very helpful if you have multiple departments all claiming credit for the increased revenue.  Or denying credit when that revenue decreases.  Attribution Models assign which touch points get which percentage of the credit.  Different types of businesses will want to use different models.  Here are a few common models used today:

1. First Interaction
          The first touch point in the conversion path receives 100% of the credit for the sale.

2. Last Interaction
          The last touch point in the conversion path receives 100% of the credit for the sale.

3.Time Decay
          The touch points closest in time to the conversion get more credit than those farther in time.

4. Position Based
          The first touch point receives 40% of the credit, the middle receives 20%, and the last receives 40%.

5. Linear
          Each touch point in the conversion path receives equal credit for the sale.

        There are of course, other models that can be used and some companies will even make their own custom models.  It all just depends on the industry you are in and what you are trying to accomplish.  Choosing which one is best for your business can be tricky, but thankfully there are tools for helping you choose.  These tools allow a company to select 3 different models to compare.  The results from each of these models can then be used to pick the model that works best.

        Knowing where to "attribute" credit will help you make decisions as well as make your employees happy.  Customers make purchase decisions at different times and in different places and you need to be able to see where they are coming from.  This is where multi-channel funnels come in handy.  These help you see all of your digital marketing channels all at once and assess the effectiveness of individual marketing efforts.  Once that is done, credit can be attributed accordingly using the model you choose.  

          There are some things to be aware of in regards to attribution modeling, no matter how important it is.  One thing to keep in mind is that for most social media sites, they are the first touch but very rarely the only touch, so they often don't receive the credit they deserve.  Another thing to be aware of is the digital nature of attribution modeling.  All models are based on online analytics and does not take into account all the off-site effort that goes into marketing, which can sometimes be much more than on-site efforts.  The last thing I'll mention is the limits everyone faces with tracking conversions over multiple devices.  It happens a lot that the conversion path is split up between a few devices and each one is recorded as a different user, so be aware of that.  The moral of the story, don't get caught up in attribution modeling. 

        Some of you may be reading this and thinking that attribution modeling is yet another necessary evil, and you'd be half right.  It is necessary to have and can be very beneficial to your company, but it is not evil.  It is cheap to implement and track as long as you already have an analyst that can work with Google Analytics.  If you don't have an analyst I think it's about time you got one.  Welcome to 2013, where business decisions are based on data.



  1. Another new concept! I feel like using attribution modeling will only become more popular as social media marketing becomes more and more pervasive. I could see social media giants like Facebook or Twitter develop in-depth attribution modeling solutions to justify the effectiveness of their advertising model.

  2. Spencer- I also did my post on attribution modeling. I like your perspective on it. I also like your diagram on the conversion process, much better than the way I described it. Good work

  3. Great job in simplifying what I'm sure can be super complex. I probably just missed it in your post, is there a standard in attribution modeling, or a preferred model?