Now more than ever, attribution seems to always be at the forefront of conversations around measuring digital marketing campaigns and their effectiveness. Traditionally, the “last click” method of measurement was standard for many ecommerce sites to gauge the success of inbound marketing campaigns across all digital channels. With robust and effective attribution modelling available within Google Analytics, access to this advanced view at channel performance is now available to everyone.
Trust us – It’s easy to get overwhelmed by the multitude of attribution models and reporting available in Google Analytics. We frequently hear the question, “Where do I start?”. Here’s a guide to help answer this, as well as provide a basic “game plan” to get started with attribution modelling.
It’s time to dive deeper by opening the Model Comparison Tool located in the Conversions<Attribution menu. We can start to see here how channel performance varies based on the model chosen:
Last Non-Direct Click: This is the stock model used in all Google Analytics reporting, and is frequently confused with “last click” attribution. The major difference of Last Non-Direct Click is that it does not provide credit to Direct on a conversion path with multiple touch points. For example, if someone came to the site initially through Organic Search and then purchased later through Direct, 100% of the credit would be applied to Organic Search.
Last Interaction (aka last click): This provides 100% of the credit to the channel that drove the visit that resulted in the conversion, including Direct. A Last Interaction model may be useful for a site that has a very short path to purchase.
First Interaction: 100% of credit goes to the channel that resulted in the first visit to the site. Let’s say someone was introduced to the site by clicking on a non-brand Paid Search ad, then signed up for Email and finally purchased through that channel. If using First Interaction, this model would assign all the credit to Paid Search. The model works well to quantify the full value of introducer channels and tactics. Try comparing the First Interaction model to Last Interaction for channels like Organic Search or a Non-Brand PPC campaign.
Linear: This assigns the conversion value evenly across all channels on the purchase path. This is a good model to use if your customer interacts with your website through many marketing channels before making a purchase. Note that this model is normally used for industries/verticals with a long sales cycle and may not be the best for an ecommerce or B2C company.
Time Decay: This model weighs channels closest to the time of conversion the heaviest, and a significantly lesser weight the longer away from the conversion the interaction was. It is also commonly referred to as “Exponential Decay” or “Time Delay”. We find this model to be very insightful and it requires little to no customization in order to get accurate results. Note that the half-life on the Time Decay model is completely customizable through the interface, and can be changed by clicking the “copy” button in the dropdown.
Position Based: This is somewhat similar to the Time Decay, but it assigns weight based on location in the conversion path as opposed to time. The default setting assigns 40% of the credit to the introducing channel, 20% of credit to influencing channels, and 40% of the credit to the closing channel. However, this is easily customizable through the interface by clicking the “copy” button in the dropdown next to the Position Based model.
We hope this helps you get started on reporting and in picking the model that makes the most sense for your business.
