Attribution Modelling & Measuring Assisted Conversions

Attribution Modelling & Assisted Conversions

What is Attribution Modelling?

When you see the channel report in your Google Analytics and review how much revenue or leads have been received by each channel, it doesn’t entirely show you the complete picture. A user may have viewed or came across more than one channel before completing a conversion on the website.

For example, a potential user views your product on Google Shopping ad and clicks it to visit the website. But he is busy at the moment and decides to come back later and not make the purchase at that time. Later that day he makes another search for your brand name on Google and clicks on your Organic search result. This time the user makes the purchase and Google Analytics credits the revenue against the ‘Organic Search’ channel. In this instance, Shopping ad did initiate the first impression for your website but Google Analytics (by default) uses the last click attribution and only the channel which led the user to the website and completed the purchase at the same time receives the credit.

With attribution models you can compare the performance for different channels in your Google Analytics account and understand their roles in contributing to conversion to your website or initiating a touch point from the potential customer during their journey of making the conversion.

Different types of attribution models

Google offers different types of attribution models and to explain them we will be interpreting them using the example below;

Tania wants to purchase a watch for her mum on mother’s day. 15 days before the mother’ day she makes a search on Google and sees websites promoting the Mother’s day gifts in Shopping ads. She clicks on your website and comes across a shawl that she likes, though Mother’s day is still far away and you offer next day delivery, she leaves the website to purchase closer to the date. 10 days before Mother’ day, she visits your website using Organic search to show the shawl to her sister that she is planning to purchase for mum. 3 days before Mother's day, she remembers that she still hasn’t purchased the shawl and needs to do it quickly. This time she has forgotten your brand name, so she does a quick search again and clicks on your ‘Search campaign’ that you have set up specifically for ‘Mother’s day’ gift ideas. She visits the website and makes the purchase.

• Last Click : As the name suggests, in this model the credit is granted to the channel that has initiated the last click which led the conversion to the website. This is the default attribution model enabled in your Google Analytics. For the above example, only ‘Search Campaign’ will receive the credit for the purchase and $100 value will be added against the Paid Search channel.
• First Click : In this model, the credit is given to the channel that has initiated the first click for the potential user. For the example above, the credit will be associated with the ‘Shopping Campaign’ and $100 revenue will be added against it.
• Linear : In this model, the credit is shared between all channels which the potential user has engaged prior to making the purchase. In our example, Tania came across three channels, ‘Shopping Campaign’, ‘Organic Search’ and ‘Search Campaign’. All three will receive the credit and total revenue will be divided amongst them i.e. $33.33 each.
• Position based : For this model, the credit is associated 40% for the first interaction, 40% for the last interaction and 20% for all channels in between. Using the model in Tania’s example, ‘Shopping campaign’ and ‘Search Campaign’ would get $40 each and ‘Organic search’ would get $20.
• Time Decay : This model gives more credit to the interaction that happened closer to the conversion. For our example, Search campaign will get most share for the credit followed by Organic search and then Shopping campaign
• Data Driven : For this model, Google uses ‘machine learning’ to understand conversions based on the past data in the account. This model is only available in limited accounts which have enough data.