Latest news

The future of attribution is data-driven

In digital performance advertising, most advertisers will report on a last touch/click model – this means that a conversion is attributed to the last piece of digital media an individual interacted/clicked on. This model is prevalent in the industry because it is an easy way to understand the impact of media on conversions. However, this loses sight of the fact that people interact with multiple media touchpoints before they convert. Many experienced marketers will have learned that if you switch off all other media, your PPC conversions will slowly dry up. Over-reliance on last click is a trap that digital advertising has repeatedly failed to escape from. Attribution models offer us a way to assign value to those other touchpoints – recognising that without them, the conversion would not have occurred, even though it was the last touchpoint before the conversion. 

Rules-based attribution models such as first touch and U-shaped still offer some bias, as we are enforcing our own opinions on the value of each touchpoint into how we report. DDA is the most sophisticated form of attribution – it uses machine learning to ingest data and assign those values and the more data that feeds in, the more accurate it gets. 

What are the pros and cons? 

At a basic level, machine learning removes human bias from the analysis and theoretically should provide a more accurate reflection of what each touchpoint contributed to the final conversion. It can also utilise modelling to make up for the reduced accuracy of cookie data. However, this is just another Google black box, and many agencies simply don’t trust the integrity of Google – we don’t know what bias they’ve written into the algorithm, and therefore, a leap of faith is required to just accept results reported on by DDA. We also encountered the pitfalls of modelling when Google introduced modelling of safari conversion in SA360 – if there isn’t enough historical data fed into the model, the output is far less reliable and has led to reported numbers that are wildly different from client databases. 

What can we and can’t we safely assume? 

  • The announcement states that DDA will be the default for reporting, with the other models (including last click) still being available. We can safely assume that once the rollout is complete, they will begin phasing out the other models. 
  • The announcement isn’t too clear on the Google product this will be available in. Still, we can safely assume that this is steadily going to roll out to all the products in preparation for their removal of cookies in 2023. This is likely Google laying the groundwork to provide a less “intrusive” method of reporting on digital performance. 
  • The adserver arm of Google products has been falling out of fashion with various browser updates since GDPR – for some advertisers, the data lost outweighs the benefits gained from using the adserver, so it makes financial sense not to use one. This is potentially Google’s first step towards claiming back that territory. 

What this means for clients  

Google’s DDA is not inherently a bad tool, and clients with enough data can see improved performance from using DDA data in their search bid strategies. And indeed, a method that can help show the value of display advertising isn’t something to be ignored. However, this announcement feels like yet another step towards the tech giants removing more control from advertisers and asking us to just trust them. Unfortunately, this ask is becoming more and more challenging.  

In the short term, there will be no impact on our campaigns as we will be able to report on last click and DDA. However, once the DDA default begins appearing in accounts, our digital teams will start monitoring the reporting compared to the other models so we can make informed decisions about the reporting when it inevitably changes in 2022. 

If anything, this is another reminder that we should all be testing and learning across other platforms to diversify our campaigns and reduce our reliance on big tech companies. 

Last week, Google announced updates to its reporting and attribution that have had digital marketers in a bit of a tizzy….again! Under the guise of protecting consumers, Google have said that from October that they will be rolling out data driven attribution (DDA) to all ad accounts with the aim of it being the standard by 2022. They have also removed the minimum data requirements to use DDA and will support more conversion types in the modelling. 

This week, Digital Account Director Gabby Krite talks through what DAA is, the pros and cons of it and what this update means for our clients. 

By Gabby Krite, Digital Account Director