By Todd Crawford on October 30, 2012
What is Attribution Modeling?
If the expressions we’ve seen on the faces of conference attendees at the mention of “attribution” are any indication, the topic is both fascinating and frustrating for marketers. Let’s face it – attribution is a dense subject. A lot goes into it.
Without oversimplifying, let’s take a quick look at attribution modeling, the benefits of multi-channel attribution modeling, and a few basic models to get started.
For marketers, attribution modeling is a set of rules which determine how to credit marketing channels for a sale, conversion, download, lead, etc. It’s the set of rules that will power their attribution reporting. First or last click referrals have been the go-to in attribution modeling for quite some time. Easy to understand and easy to implement, these models give full credit to either the first or last channel a consumer interacts with before completing the conversion.
The latest trends in attribution favor multi-channel models in which multiple or all marketing touch points prior to a conversion are taken into consideration and allocated a portion of the credit. A multi-channel attribution model can be set up in any number of ways depending on the number of channels in the mix, the time frame for which a touch point is considered to influence consumer behavior, the order in which consumers encounter each channel… the possibilities are endless. And the more factors you begin to take into consideration, the more complicated the model becomes.
As a marketer, you probably understand that consumers are normally influenced by more than just the last ad they see before making a purchase. On a consumer’s journey from first hearing about your brand to conversion, they might encounter a TV ad, a search result, a banner ad online, or any other number of your marketing initiatives.
The purpose behind multi-channel attribution modeling is to give you a clear understanding of how a consumer’s interaction with each channel influence their actions. Rather than attributing a conversion entirely to a single channel, the multi-channel approach helps marketers determine the value of each channel in relation to their conversion goal. This allows marketers to better evaluate the resources allocated to each channel. For example, in the performance marketing arena, attribution modeling helps properly determine payout for partners based on the degree of influence they had on a conversion.
A Few Models to Get Started
While creating your own custom attribution model may seem to be the most desirable, it most likely will not be easy. There are several popular multi-channel attribution models you can use to get started and guide you towards your own preferred model.
- The Linear/Equal Attribution Model – This model assigns equal credit to every touch point leading up to a conversion. If a consumer touched 4 different channels before purchasing, each channel would be given credit for 25% of that sale.
- The Time Decay Model – In this model, touch points are assigned credit based upon closeness to a sale; the last interaction is assigned the greatest percentage and the first would receive the least. So, for example, if a customer clicked on a banner ad the day they converted and had also opened an email two weeks prior, the majority of credit would be given to the banner ad.
- The Position Based Model – This model assumes that the most influential touch points are the first and last, assigning them equal credit with the touch point in the middle receiving the least. Assuming there were 5 interactions with your marketing efforts, this model could give 30% credit to the first and last interactions, 15% to the second and fourth, and 10% to the middle.
While determining which attribution model to use (or create), look at the information you already have on your consumers’ journey to conversion. Look for patterns in touch point progression. See how many touch points are crossed on average. And, as you see the results come in, test the results by reallocating your investments in different channels based on their performance to see what impact it may have on conversion rates.