Cookies

The role for AI-driven, cookieless attribution as third-party cookies die

12th April 2024
Carly Bell

With 83% of marketers indicating that they are reliant on third-party cookies to support effective marketing – and a Google ‘switch off’ that is now underway – finding alternatives is high on every marketer’s agenda right now. 

For us, cookieless attribution has always made perfect sense. Here we explore a bit more of the reasons why including: 

Why Google has decided to remove third-party cookies?

This is a topic we have covered in quite a bit of detail in Part 1 of our Cookie Countdown series but in a nutshell, third-party cookies are going away for a combination of reasons including: 

  • Privacy concerns – both individuals and privacy campaigners have become increasingly concerned about the way that the marketing and advertising industry collects and processes personal data
  • The need for Google to act – as it lags well behind other browser market players like Apple (Safari) and Mozilla (Firefox) in terms of action on the privacy front
  • Scrutiny from legislators – who are increasingly not just legislating on privacy but handing out sizeable fines to players like Google too.  
  • Ad-blocking is on the increase – users are also increasingly taking things into their own hands with 63% installing ad-blockers 

Google began the process of deprecating cookies for Chrome users in January 2024. However, despite Google having proposed potential cookie replacements as part of their Google Privacy Sandbox, the industry response has been muted at best and marketers are concerned – with our research indicating that 83% of marketers are reliant on them. 

The situation has led to many marketers considering alternatives to traditional cookie-based measurement – one of which is AI-driven, cookieless attribution. 

What is cookieless attribution – and how can it help post-cookies? 

In this section, we take a closer look at why traditional models of attribution are failing and how AI-driven, cookieless attribution can plug the gap. 

While the need to make plans for the end of third-party cookies feels like a relatively recent issue for marketers (leaving aside the various delays in the process to date) we have a longer history of exploring the limitations inherent in cookies – and the analytics and attribution solutions that rely on them. 

One of the early learnings on our journey to find a better way to perform effective attribution for our customers was the fact that third-party cookies are not good at tracking the type of complex, multi-device journeys that buyers are typically on these days. And, if you stop to think about it, it stands to reason. 

Take programmatic advertising as a prime example. It is a channel supposedly fuelled by the highly accurate third-party data generated by third-party cookies but the reality is that programmatic has singularly failed to deliver on its ‘right ad to the right person at the right time’ mantra – as evidenced by research that shows you need to spend 2.5 times the cost of a display ad and 1.5 times the cost of a video ad to gain any sort of incrementality whatsoever.    

Source: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3203131

So why is that? 

Part of the answer lies in the combination of the particularly poor job that third-party cookies do of tracking the impact of every touchpoint on the customer journey (due to limitations in the cookie/pixel tracking approach being used) and the use of over-simplistic attribution models which are inherent in cookie-based solutions like GA and Adobe. Compare the practical example below where the user has been impacted by a range of media touchpoints from an early Paid Social click-through to a late funnel Paid Search click-through. 

A cookie-based solution will attribute 100% of the conversion value to either the Paid Social or Paid Search interaction – depending on whether you were using First or Last-Click attribution. Which is a gross over-simplification in anyone’s language. Our own AI-driven attribution solution can take into account, and accurately value, the impact of ALL touchpoints on the journey. 

Rebuilding your broken marketing data using Machine Learning is essential

One of the key implications in the background here, and a knock-on effect of poor tracking of these journeys, is that the underlying data generated by solutions like GA and Adobe is effectively ‘broken’. 

Why does this matter? The example below illustrates the reason perfectly. 

Here we worked with leading UK retailer Tesco to rebuild their broken GA360 data in our own Corvidae attribution platform – and were able to establish that 80% of it was incorrectly attributed by the cookie-based measurement solution.  

With huge implications for the quality of their attribution based on the data. 

So how exactly are we able to effectively ‘fix’ the broken data that has been generated by the cookie-based solution (in the case above GA)?

Our Corvidae platform can collect and ingest data from your existing advertising and analytics platforms and join it to your own website’s clickstream data. From there Corvidae can use AI and Machine-Learning techniques to effectively stitch together disconnected user journeys to provide much longer journeys that take into account the impact of each touchpoint on the way. From typically hard-to-measure early funnel interactions on display and social through to lower funnel activity like PPC and re-targeting. 

Stitching longer journeys together for more effective attribution

In the process outlined above, AI is effectively replacing the measurement role previously performed by third-party cookies. But doing a much more effective job of producing a single, unified journey that connects all touchpoints to conversion. 

And crucially in a way that is not impacted as the user switches devices – which is a major problem in cookie-based solutions.

This AI stitching is a unique globally-owned patent that is owned by us. 

 Driving down CPAs using AI-driven attribution

 So, how does this improved view of customer journeys help marketers be more effective? 

Put simply, using AI for attribution lets you reach up the funnel and identify lower-cost CPA opportunities in earlier phases (for example, in the ‘See; and ‘Think’ phases of the See Think Do Care model) that are being completely undervalued due to limitations in standard cookie based measurement systems. 

The “Do” phase in this model – where users show intent to buy – is characterised by high levels of competition for limited advertising space which is resulting in grossly inflated CPAs. This has been made worse by other factors including the pandemic (where CPAs increased by as much as 89% according to Forbes and have remained high) and the focus on lower-funnel attribution by the big advertising platform players like Google Ads and Facebook ads – who continue to measure based on the limited customer journey view that their cookie-based solutions have. 

AI-driven, cookieless attribution lets you stop off the CPA treadmill by stepping into ad auctions in the See and Think phase that are much less competitive – with fewer advertisers competing for space. 

Automating Google Ads using longer journeys

The good news is that it is possible to use Corvidae to effectively give platforms like GA the information they need for their AI to optimise the bidding process and drive down your CPA levels. 

This is the type of approach we used for a leading Electronics retailer in the UK/EU & North America which saw CPA reductions of 15.6% and an increase in clicks of 10.3%. 

AI-driven attribution in action – reducing CPA by 46%

This is the approach we took with a large telesales company in Germany. 

This company is a multinational corporation specialising in televised home shopping. It broadcasts in seven major countries to 350 million consumers.

The organisation were looking for an attribution solution to replace their existing ‘Last-Click’ reporting in a GDPR-compliant manner – and, with several analytics solutions in play including Adobe Analytics and a customer internal reporting system – it was essential for them to see the true customer journey.

Using Corvidae we were able to effectively rebuild their marketing data, and generate journey lengths that were 3 times longer than before, to better inform spending decisions which led to the: 

  • Customer acquisition cost was reduced by 46% on Google Ads
  • An increase in ROAS of 86%
  • A reduction in CPC of 34%
  • a 3% uplift in Net New customers (for the same level of sales volume)

Delivering game-changing cookieless attribution with Corvidae

Corvidae is our patented, AI-driven attribution solution and it enables you to:   

  • Rebuild your marketing data – to provide an accurate picture e of what is driving conversions across your marketing mix
  • Reduce the cost of your CPAs – by identifying lower-cost, higher funnel opportunities in media like display and paid social
  • Improve ROAS in existing channels like Google Ads – by automating the process of optimising your campaigns using the AI built-in to their offering

Don’t believe us? Why not see for yourself! 

Need to learn more first? Download a copy of our ebook – Is Cookie-Free Attribution a Myth? – and find out why AI and Machine Learning are key to post-cookie success. 

Related Content

Is Cookie-Free Attribution a Myth?

Download our eBook