Attribution in Digital Marketing – Digital Analytics Blog

Attribution analysis enables marketeers to understand how each marketing channel has contributed to the sales effort. Attribution analysis is a crucial tool for any analysts who want to understand their customers and optimise the ROI of their marketing campaigns. In this article we’ll look at how it works and how to use it correctly to leverage the maximum value. 

What is Attribution in Digital Marketing? 

Simply put, Attribution is a method used in Digital Marketing to identify how much each marketing channel has contributed to your sales efforts. It seems quite straightforward on the surface, but when you dig into the topic, it can rapidly become complex. However, with the right processes and methodology, a properly functioning Attribution analysis can significantly help you to optimise your marketing ROI. 

Why use Attribution analysis? 

Today, online consumers interact with brands in increasingly divers ways. Technology provides an ever expanding number of channels to push the right brand message to the right user when they are most receptive – and on an increasing number of devices. When a user converts, it is also more and more complex to understand which interactions between a brand and a user have truly had an impact.   

Recreating consumer journeys and deciphering this type of information can be extremely complicated. Aside from data quality issues and the reliability of third-party cookies, consumer journey datasets are quite complex.  

• How far back should you aim to reconstruct paths?
• Which interactions should you take into account?
• Which should you exclude?
• Are your ad networks providing you with reliable and complementary datasets?

It can get tricky, quickly, and I’m not even going to mention the potential biases of third-party Attribution providers when they have stakes in various ad networks. 

Nevertheless, when approaching this topic with a scientific and methodological approach – and by removing the external factors you can control – it is entirely possible to get a clear picture of what is happening behind the scenes. Combined with computational power and some economies of scale, an accurate attribution analysis with the correct interpretation can provide you with the right answers and insights. This will ultimately help you optimise the ROI for your marketing initiatives. 

What are the different Attribution models? 

Before going into what attribution can do for you, here’s a definition:  

Attribution in online marketing is the process that credits conversions to the correct marketing activity, or combination of activities. Attribution analysis provides an in-depth study of marketing activity performance over a chosen analysis period, against an attribution model of your choice. 

The first step to Attribution is to recreate customer journeys – and for this you need a 360° view of your consumers. You need to measure all the interactions between consumers and your brand across multiple devices and across all your various platforms, and then organise them in chronological order. 

Once you understand the journeys taken by users from brand discovery and awareness, through to consideration-to-purchase, you can apply a range of attribution models. This attaches weight to the different interactions based on their contribution towards each step consumer journey.  

There are two broad categories of attribution models each with their own advantages and disadvantages – Rule-based attribution models and Algorithmic models. 

Rule based Attribution Models 

Rule-based attribution models have the advantage of being cheap, easy to implement and are completely unbiased. This category can be divided into the following models: First click, Last click, Linear, Time-decay and U-shaped. 

• First click – the first touchpoint obtains 100% of the credit for a conversion. 
• Last click – the last touchpoint obtains 100% of the credit for a conversion. 
• Linear (uniform) – the credit is shared evenly between the various touchpoints. 
• U-shaped – the credit is shared between all touchpoints: X% for the first touchpoint, X% for the last touchpoint, and the rest is shared between all other touchpoints
• Time-decay – the credit is shared amongst all the touchpoints, with weights increasing linearly the closer in time a touchpoint is to the sale

There is also a subset of rule-based attribution models: custom rule-based models. Here the analyst defines which weights to apply to each interaction based on how it is positioned in the consumer journey. This method has the advantage of tailoring rule-based models to the business’ strategy. However, the model is less impartial and takes some prior knowledge of consumer journeys to setup. 

Algorithmic Attribution Models 

This category of models uses advanced data science and machine learning techniques to identify tailored weights to apply to each interaction that led to a sale, based on a wide variety of factors. These models provide an unbiased approach to attribute credit for sales and can provide considerable results with economies of scale. However, there are a few drawbacks… These types of models require large amounts of data, time, and development for them to be accurate. In addition, they require advanced knowledge in data science to set up and can prove to be a costly investment to develop and maintain, hindering their main goal: aiding in optimising marketing ROI. 

How do you define your Attribution strategy? 

Now that we know how attribution works as well as the pros and cons of the various attribution models, you can focus on which one you should apply to your business. 

There is no right or wrong attribution model; they will all provide a valuable perspective to understand consumer behaviour and optimise marketing ROI. The key is in selecting the optimum models – those that most accurately reflect your business strategy… 

Are you currently focused on user acquisition and brand awareness?  
→ Analyse your data via the lens of a first click model.  
 
Are you having trouble converting potential customers after they are exposed to your brand?
→ Look at using a last click model.  
 
Do you have long consumer journeys from discovery to purchase and every touchpoint is critical?
→ Look at using a linear or time-decay attribution model. 

 
Once you have identified the relevant attribution models to your business, you need to stick with them and benchmark performance over time to measure improvements in your optimisations. 

As well as using multiple perspectives and benchmarking to analyse your progression, it’s also important to include human validation to contextualise the results and ensure your data makes sense. This also enables you to understand the value of each touchpoint and how it fits in with the other interactions with your customers. 

AT Internet’s Analytics Suite Delta has a natively integrated attribution solution that builds on the quality data we already collect. It aims to provide all the information you need to explore user journeys and better understand your consumers’ behaviour. With its model comparison feature, you can analyse, compare and benchmark your performance with multiple perspectives at the same time – all at the click of a button.  

Click here a free demo and find out how you can leverage user-centric digital analytics to drive the value of your online offering.