It is not enough to say the last ad channel the user saw before converting was the sole reason for buying and therefore the cost per acquisition was simply the cost of that ad airing (the Single Touch model). For each user journey to conversion there are likely to be several channels, ads and site mentions seen by the user before his/her enters the website and converts, including remarketing advertising after the user leaves.
Each of these ads and site mentions needs to have a value assigned so that the cost per acquisition metric calculations are more accurate, ROI is correctly measured and marketers have the right data to decide going forward the opimal mix of marketing activitiy to maximise revenue and future growth.
So let's explore Attribution Models in more detail and look at recent developments in this field.
Single versus Multi Touch Attribution Models (MTAs)
Single Touch Attribution Models simply give 100% of the credit for a conversion either on first click or last click, or sometimes the mid click in the user journey. Multi Touch looks at more of the journey of ads the user experiences. But different models of Multi Touch assign values differently:
- Linear attribution: Gives equal credit to each and every touchpoint
- Time-decay attribution: Assigns more credit to touchpoints closer to conversion
- U-shaped (position-based) attribution: Emphasizes first and last touchpoints
- W-shaped attribution: Highlights first, middle, and last touchpoints
- Algorithmic (data-driven) attribution: Uses machine learning to determine credit distribution
Which MTA to use?
It's an interesting question, and will ultimately be driven by the nature of the user's behaviour and the products/services being offered by the company.
But how do marketeers know accurately which model to use?
With the growth of AI and improvements to data processing techniques, more are turning to the Algorithmic model option, using data processing techniques to understand what is driving the biggest conversion successes and the optimum mix of ad creatives, spends, mixes of channels etc.
The issue of cookies
MTA models rely on cookies to collect data but with growing restrictions on 3rd party cookies the accuracy of tracking user journeys is getting more difficult. Google are phasing out 3rd party cookies on Chrome browsers and Safari already has.
There's another issue with MTA models - they only look at the online journey and ignore offline (posters, TV ads etc) and word of mouth.
Enter Marketing Mix Models (MMM)
A model favoured by several top digital marketing agency companies in the UK, MMM type models look at the data in a top down fashion. Instead of applying a certain type of MTA model and all the assumptions that come with it in terms of weighting, MMMs simply correlate results to the many different combinations using multiple regression techniques.
In doing this, MMMs don't assign weightings to touchpoints, the matchs does it for you. Climate control experts using MMMs to predict the global warming levels we are now seeing.
Another benefit is quite fundamental. MTAs essentially are backward looking and are therefore weaker than MMMs, which look forward, in predicting.
Time lags is another area to look at. Some campaigns can take much longer to show their impact. Brand awareness campaigns are a good example; these top of funnal campaigns are are less about immediate conversions and more about longer term decision making. Some people may repond later to a brand campaign and completely miss the other more immediate digital campaigns and be missed off the measurements of ROI. An MMM approach would be more likely to pick these up.
Attribution Platforms and Softwares
Some of the platforms include Google's own GA4, Adobe Analytics and Mix Panel. Software options include Amazon Kinesis, Proof Analytics and SuperMetrics. More information can be found on this article from digital marketing agency Jam.