Knowing how different marketing channels work together to drive conversions and sales is key for marketers, and there are several attribution models available to help you examine your data from one angle or another.
But that array of options often creates more confusion, rather than insights. Which attribution models give marketers the best picture? That’s the question we’ll tackle in this post, by analysing the pros and cons of the 6 main attribution models.
An attribution model is a set of rules an analytics platform uses to map conversions and sales to respective touchpoints in a customer’s journey. People come to your website through different channels:
Attribution modelling shows you which ones bring in the most sales and which ones assist conversions – the marketing actions/channels that contributed to sealing the deal. But being a descriptive analytics tool, GA can only tell you how users interact with your website and other properties before converting.
It prescribes the right actions or predicts how conversions will change if you optimise a certain channel – it’s your job to properly capture and interpret all the available data. Arguably, that’s the hardest part of any marketer’s job. But that’s where data science can help.
There are several attribution models you can consider:
The credit for sale (conversion) is assigned to the last platform or channel that a user came from before converting.
Example: you are hosting a webinar and place a link to your product in the description box. Out of 100 attendants, 10 clicked that link and signed up for your tool. According to the last interaction attribution model, 10 sales will be credited straight to the webinar platform (referral traffic).
Sensing a fallacy here? This model does not account for all the other marketing activities you’ve used to generate the buzz around your webinar – that aggressive email campaign you did; the paid social media promotion and all the blog posts you’ve published to gather a warmed up, relevant audience for your event. The last attribution model does not account for all the other touchpoints a user probably had before converting.
All the credit goes to the first touchpoint a customer had with your brand. You assume that once the person is there, nothing else can push them towards or derail from converting.
Example: Tim googled your blog post. Two weeks later, he clicked a display ad or typed in the website URL directly and placed an order. Organic search will get full credit. Other actions are not considered.
This one accounts for every touchpoint a customer had with your business before converting. All the involved channels receive equal credit for each sale (e.g. 50%-50% or 25%-25%-25%-25%).
Example: Jane takes the time to interact with your brand before converting. She stumbles on your blog post (organic 25%) + clicks on your tweets (social media 25%) + checks your email newsletter (25%) + types your website URL directly and buys a subscription (Direct 25%).
In this case, different touchpoints get fairer “weights” based on their position in the user’s journey. The first and last touch points receive a 40% credit allocation as they are deemed more important than what happened in the interim.
Example: Leo found your brand on Instagram (40% social media). He subscribed to your newsletter and clicked a link (20% email). Later he placed an order through an affiliate link shared by his favourite blogger (40% affiliate).
Touchpoints taking place closest in time to sale/conversion receive the most credit. In other words, the first click has the least value, the last one reigns supreme. But every action in between is measured as well. For additional accuracy, you can add a half-life for a certain decay. The touchpoint at that period will receive 50% of the credit of that final touchpoint.
Example: Lisa navigated to your website via a search ad (10%). Later she found your website again by googling it (15%), subscribed to your newsletter, and clicked back just in time for sales (25%). But her shoe size was not in stock, so she came back directly a few days later (50%).
The proposed attribution models can be further customised with additional rules and benchmarked against one another, for example by using the Google Model Comparison Tool. Custom modelling gives you more room for reconciling the analytics with the actual customer journeys. However, it will still assign the same predetermined weights to different channels.
So if every model is flawed in one way or another and does not provide a full picture, what’s a business to do? The answer is to use a data-driven attribution model.
Unlike others, this model leverages actual data from your account to generate a custom model, personalised to your business, for determining various touchpoints throughout the entire customer journey and assigning custom credit weights to them.
This all sounds great but, unfortunately, Google only offers it to Google Analytics 360 customers. To get this, there is a charge of $150,000 (circa £115,000) per year and you must meet the following criteria:
If you don’t have the budget to pay for Google Analytics 360 or don’t meet the criteria, fear not. It is still possible to use a data-driven attribution model – find out here in part 2.
George is an SEO Specialist and Data Scientist in the Portsmouth office. He has worked in E-Commerce and Digital Marketing across many industries, for and with companies all over Europe. Before joining Vertical Leap, George worked as Marketing Director for his own company, for which he managed to expand the company’s activities to 5 European countries. George has been creating websites for more than 10 years and he has in depth experience in designing and bringing optimised E-Commerce websites to market.
Categories: Data & Analytics, PPC, SEO
Categories: Data & Analytics, PPC
Categories: Data Science