The customer journeys can be long and complex. Fortunately, we can map and trace them using analytics methods like attribution modelling. With this tool, you can better understand your customers’ behave and allocate your marketing budget more effectively. It is therefore worth to know how attribution models work for successful online advertising and achieving good results.
Attribution models – definition
The official definition of attribution models is a bit tricky. They are the rules that determine how credit for sales and conversions is assigned to touchpoints in conversion paths. It doesn’t sound easy, right? So let’s go through it again, but this time using simpler words.
We will first explain what is conversion – after all, we find this notion twice in attribution models’ definition. The conversion can be made in different ways – it isn’t just a product or service sale, but all interactions with an online store, i.e. clicks on Google ads or remarketing banners.
We use attribution models to measure the impact of each action on sales. With that method, we can more specifically analyse the customer journey, which very rare leads straight to conversion. Before customers make the final decision, they usually visit the store page many times, seeing and interacting with ads in the meantime.
Last click – the most common attribution model
Currently both – Google and Facebook – automatically assign conversion credit based on the last click attribution model, totally ignoring other actions made by customers before buying a product or service.
However “last click” is not the only available attribution model. There are also 5 other types of it. In the following paragraphs, we explain how they work.
Other attribution models available in Google Analytics
In this model, the first click that lead visitor to the shopping path receives 100% of the credit for the conversion.
If we assume that our customer clicked on store’s ads, that were visible in many places and associated with many keywords, 5 times and the at the fifth time, they decided to buy the product, the first click would receive 100% of the credit for the conversion.
In this model all interactions on the path share equal credit for the conversion.
Assuming that our customer clicked on store’s ads, that were visible in many places and associated with many keywords, 5 times and the at the fifth time, they decided to buy the product, each interaction would receive 20% of the credit for the conversion.
Choosing this model you give more credit to ad interactions that took place just before the sale had happened. Other ad interactions also gain credit for the conversion but much smaller.
Assuming that our customer clicked on store’s ads, that were displayed in the course of 8 days in many places and associated with many keywords, 5 times and the at the fifth time, they decided to buy the product, the interactions closest to the conversion would receive 50% percent more credit than the earlier interactions.
In this model the first click that leads the visitor to a shopping path receives 40% of the credit for the conversion, and the last interaction with an ad gets another 40%. The remaining 20% is assigned to the rest of the interactions.
Assuming that our customer clicked on store’s ads, that were visible in many places and associated with many keywords, 5 times and the at the fifth time, they decided to buy the product, the first and the last click would receive 80% of the credit for the conversion, while the all interactions in between – 20%.
This model is only available to accounts with enough data, which are necessary for tracking conversions along the customer journey. It is the most accurate model, perfect for big e-commerce websites.
How to choose the right attribution model?
Google experts suggest choosing the data-driven model whenever possible, because analyses obtained through this method are the most precise.
However, we don’t always have enough conversions to use this model. In that case, it is more suitable to choose another option adjusted to our marketing goals and business type.
The model that you choose should reflect the profits that you earn by advertising in particular online channels. You should be able to verify whether all your campaigns are truly effective. If not, maybe it would be better to cancel one of them and spend the money on a different advertising channel?
Example scenarios for using some attribution models
If you have many visitors on your site and you want to increase the sale of the products, you should choose one of the conservative models, which concentrate on latest interactions.
The conservative models are:
- Last Click (the most conservative).
- Time Decay.
If you are developing your online business and are searching for clients, you should choose one of the growth-oriented models that help you find the places where the customer journey begins.
The growth-oriented models are:
- First click (most growth-oriented).
- Position based.
If you advertise in many different channels and want to check their efficiency, you should choose the linear model, which helps you verify the efficiency of each campaign.
Remember – there is no perfect attribution model. It is impossible to choose one that will be suitable for every business. You should draw conclusions yourself through testing and comparing different models.
How to find the time for attribution modelling?
Finding the right attribution model could take a lot of time. That’s why many marketers choose the most common, but as we already mentioned less accurate, last click attribution model.
What If you could save the time spent on preparing product feeds for omnichannel campaigns and use it for detailed analysis of customer journey behaviour?
With Feedink you can create a product feed that complies with technical specifications of many platforms with only one click. All your data will be stored in one place and automatically updated. Thanks to that your adverts will always have actual prices and descriptions. Automating the product feed management this way, you can save enough time to be able to concentrate on other marketing activities, like attribution modelling.