Attribution is the method used to give value to each of the touch points in the customer journey. This journey is becoming more and more complicated and fragmented between devices, traditional medial like TV and radio, social media, online search and physical interactions. As the good old saying goes: “I know only 10% of my marketing works, but I don’t know which 10%”
Below is my take on explaining attribution models with Monopoly tokens. I thought that for some of you, this might be more productive than reading a whole article 🙂
The subject of attribution is often neglected amongst marketers – it is dry, complicated and it is pushing people out of their comfort zone. It often brings up uncomfortable truths and exposes internal problems like tracking, collecting data and distributing budgets.
Why is Attribution important?
If you are not allocating value to certain parts of the customer journey you are missing the full picture. People nowadays are bombarded with advertising online and offline. Their attention span is short – they are impatient, curious and demanding. The use of multiple devices is very common. Advertisers usually try a mix of channels to reach them – some traditional and some new. However, the way we track and attribute value along the journey is very traditional.
Most advertisers still use last click when it comes to online attribution. This means that the channel which took brought the last click before a conversion takes all the credit. This approach paints a very misleading picture of the reality and leads to wrong decisions.
What types of attribution exist?
Besides last click there are few other rule-based attribution models. They are called rule-based, because they are static – there is a rule which is always applied over, no matter what the case.
Linear model
This model distributes the value equally across every point of the customer journey online. It could still be misleading, as not all touch points are equally important.
Position-based
This is a model which gives 80% of the value to the first and the last points of interaction and spreads the rest equally between the other touch points in between. Still this model dives quite a bit of credit to the last touch click.
Time decay
This model values most the last touch point and everything before it has lesser and lesser value in the journey. This is a slow transition from last-click attribution. It still gives some information about the previous interactions, but the value of the initial touch point is the smallest.
First Interaction
This model gives the full credit to the first touch point and ignores all touch points after that. It is a bit of a drastic model to use, but it depends on what type of campaigns you are running. If tend to get conversions quickly after the first interaction, this model might work for you. You must still have in mind that this model hides a lot of important interactions along the way to the conversion.
Last Non-Direct and Last Google Ads Click
These models are giving the credit either to the last interaction before a direct search (which is usually the last one in the path), or the last Google Ads click. The rationale is that you would like to see which last paid Google click brought the conversion. The problem with this model is that the last Google Ads clicks tends to be from a branded search campaign. This model will undervalue your non-branded search terms and display campaigns.
As you realize each one of these models has its flaws and will not fully show the whole picture. Also, depending what type of attribution tool, you use you might only see parts of the information – Google Analytics would always attribute more value to Google Ads campaigns. Facebook’s analytics would attribute more conversions to Facebook. Third party attribution tools can be used to analyse the full picture. Typically, they cost a lot and only big companies or agencies can afford using them.
New developments in free attribution tools come from Google with their Google Attribution product which would be part of the Google Marketing Platform. This tool is advertised as a single place to track all conversions, using data-driven attribution.
What is Data-Driven Attribution (DDA)
This model is available to Google Ads advertisers who have minimum of 600 conversions and 15,000 clicks last 30 days. The model is only available for tracking conversions within Google Ads. This is the minimum amount of data for Google to use machine learning to attribute value to different touch points along the journey. The system works by assessing every step of the conversion path. It analyses patterns which tend to bring conversions over time and allocates value according to the touch points which have highest probability to bring conversions.
Comparing attribution models in Google Analytics
One way of understanding your data is to start comparing results within Google Analytics. If you navigate to Conversions and then to Multi-Channel Attribution Comparison you can easily pair each of the models (but no DDA) and compare how conversion number and cost per conversion varies between your different sources.
Google Ads tracks conversions by default with Last-click attribution model, whereas Google Analytics uses Last Non-Direct click by default. This is why it is crucial to make a comparison and see whether you are not following a wrong path by relaying too much on data from one or another platform.
How to implement an attribution model change in Google Ads?
When changing the attribution model, you must be aware that any Smart Bidding campaigns will immediately enter in learning mode. This will influence the performance of your campaigns and cause lag of conversion record. Having said that, if you want to move away from a last-click attribution, which has been the default for a long time, you will have to put up with a few weeks of performance turmoil. Therefore, you better make the change in a quiet period for the business.
You choose an attribution model for each one of your conversions. If you are tracking more than one actions as conversions on your website, it would make sense to adopt the same attribution model for all of them. You must have a strong rationale why using a different model on certain conversions.
To change the attribution model, go to Conversions and then choose the conversion action you would like to change. You can select the model you which to have from a drop-down menu.
Cross-device attribution
A cross-device conversion is when a user started their journey on one device and finished it on another leading to a conversion. If you are tracking cross-device attribution you can understand what the important of mobile, tablet and desktop devices is on your marketing. It is a very important report and it Google Analytics tracks cross-device conversion if there has been a Google Ads click on the conversion path – regardless on which device.
The Device Report in Analytics open Beta project which you can opt in. It allows Google to match the data of your website users and to show you the unique visitors. This means that if someone came to your website with few different devices, but logged in with the same account, they are registered as one user.
View-through Conversions
In Google Ads you can add a column called View-Through Conversions. These are conversions which happened after a user was shown a Google Display ad and converted in a specific time window after. The window is determined by the advertiser when setting up their conversion action. 30 days for a conversion action will result in a view-through conversion in up to 30 days after the user saw a display ad. View through conversions are not included in the total amount of conversions in Google Ads, but are automatically added to your Facebook conversions.
