Cookieless Attribution FAQs

We recently hosted a webinar around the realities of achieving cookieless attribution in a post-cookie world. If you didn’t get a chance to watch the session live, you can watch the on-demand recording here, or read our roundup blog.

During the session, we received some fantastic questions but, as we only had 20 minutes to cover a lot of ground, we didn’t get the chance to answer them all. So, I wanted to take some time to answer those unanswered questions and queries around our cookie-free attribution software.

Click on the question below to navigate to its answer:

• EU regulation last week ruled again that GA was illegal for use by businesses operating in the EU. How can Probabilistic analytics address that if at all?
• What about alternatives to third-party cookies, like the IAB Framework for compliant ID sharing?
• How do we account for the discrepancy between Advertising report KPI from agencies (i.e: sessions or traffic) vs GA reporting (i.e: sessions or user)? How do we know the advertising we are running is truly driving traffic to our websites if we can’t see the full picture in GA?
• Does Corvidae use a Markov model?
• How are you gathering clickstream data without cookies – or browser extensions?
• How are you factoring in exposure to your attribution model? Clickstream data is likely to be very limited to certain channels
• How does the AI work? Does it use first-party cookies or work another way?
• How are you determining the accuracy of your models? How transparent are you?
• What would be the benefit of AI-driven attribution versus something like MMM modelling?
• Do you utilise any third parties to build out a device graph?
• What are your thoughts on total attribution/unified measurement?

EU regulation last week ruled again that GA was illegal for use by businesses operating in the EU. How can Probabilistic analytics address that if at all?

The Schrems II ruling in 2020 was a milestone in EU policy regarding consumer data rights. The EU took the position that businesses cannot offload responsibility for data compliance onto their technology provider, and instead retain GDPR liability should EU data go outside the EU.

In February, both Austria and France have ruled that Google Analytics is illegal from a GDPR view, as it cannot prevent data passing to the US, and the US security services have provably accessed Google data. It is expected that all EU member states will fall in line with this position over the coming months.

The last few weeks have seen the first major businesses be taken to court for their use of Google Analytics.

Using Corvidae offers a GDPR compliant replacement for GA and Adobe marketing analytics, as it stores all data compliantly and securely. In addition, Corvidae has been passed for data compliance by UK banking, meaning UK and EU businesses can deploy Corvidae with confidence that they will not incur GDPR liability.

As it stands, liability exists for any GA or Adobe users who collect data from anywhere in the EU – if you receive traffic from Germany, for example, you would be liable for breach of data compliance.

Corvidae’s patented AI technology removes the need for any cookies, making it possible to collect and measure performance, compliantly, with over 95% accuracy, compared to only 20% accuracy when using cookie-based systems.

What about alternatives to third-party cookies, like the IAB Framework for compliant ID sharing?

IAB Europe’s proposed Transparency & Consent Framework (TCF), touted as a replacement for third-party cookies, has been ruled as illegal by Belgium already.

The IAB’s other infrastructure intended to support third-arty cookie replacements, the UID 2.0 (now rebranded as the IAB Tech Lab Tokenization Framework) has also hit turbulent water as the IAB has decided to announce it will not continue as the technical administrator, effectively stepping back from offering the vital infrastructure required to support a global cookie replacement which would have staggering server costs associated.

Taken together, the TCF and Tokenisation Framework will launch dead in the water from a technical infrastructure and compliance point of view, and are unlikely to offer a real alternative way to circumvent the intent of GDPR, which is to protect customer data and consent and restrict it to first-party relationships only.

First-party data compliance is built into the architecture of Corvidae, which holds a global patent on the use of AI to compliantly build first-party conversion paths that are complete and typically 90-95% accurate, representing a huge step-change in performance from first-party cookies, which only offer around 20% accuracy.

How do we account for the discrepancy between Advertising report KPI from agencies (i.e: sessions or traffic) vs GA reporting (i.e: sessions or user)? How do we know the advertising we are running is truly driving traffic to our websites if we can’t see the full picture in GA?

In short, you can’t align GA.

Your agency can only add to your understanding of performance by sharing data from the AdTech platforms they use – Google Ads, Facebook, TikTok, etc. Ultimately, those numbers will overrepresent actual impressions and revenue you are tracking in GA or Adobe.

The only way to align is to require a single source of truth using an agreed attribution solution.

Does Corvidae use a Markov model?

No.

Corvidae attributes revenue to the most granular level possible – the individual visit or impression, and then allows aggregation up to insightful segments.

Shapley and Markov models cannot handle such a large amount of data points in their operation and the underlying mathematics fail. Due to this, we have developed our own attribution modelling techniques to allow our own approach to be viable.

How are you gathering clickstream data without cookies – or browser extensions?

With a first-party pixel.

Our pixel uses an Event Stream architecture, meaning it can be deployed in any conceivable environment, including:

• website
• app
• ad impression
• and in custom middleware.

How are you factoring in exposure to your attribution model? Clickstream data is likely to be very limited to certain channels

Using our AI model.

Impressions that are not associated with a click are associated using the models assessment of the click and click behaviour and the probability that the visit would have seen the impression. This plays to the strengths of AI, and is measurable when training the AI.

We test and validate accuracy of this stitching by asking the model to classify conversion journey we feed it as converting or non-converting. We then compare it’s prediction to the true conversion state of the journey. No model is released which scores less than 85% predictive accuracy.

How does the AI work? Does it use first-party cookies or work another way?

No, it uses no cookies at all.

Our model uses AI to understand the difference between a transacting and a non-transacting path to discover which touchpoints are influencing in your customers’ decision-making process.

Our model is trained to understand each individual customer, demystifying which interactions were important to their decision to buy.

Allowing Corvidae to tune itself to your customer’s behaviour – rather than making you decide what to value in advance.

You can think of the scoring of the AI as a measurement of the incrementality of each event in the conversion path towards a conversion – or non-conversion. This is the basis of Corvidae’s attributed value.

How are you determining the accuracy of your models? How transparent are you?

We use anonymised test and train sets from our customers’ historic data to build and validate our models.

These sets are made up of thousands of transacting and non-transacting journeys.

The transacting journeys have transaction events removed.

Given these user journeys (impressions, click throughs and visit hits) we create a vector-based model that predicts whether a transaction would happen or not.

This is then compared to the complete journeys, including transactions, to test how accurate the model is at predicting transactions, and hence judging the engagement of an individual user at different points in their user journey.

Given that the model can weigh each event on how important it is for the transaction or not, that weighting would be used as part of the attributed share.

What would be the benefit of AI-driven attribution versus something like MMM modelling?

Media Mix Modelling (MMM) is complementary to Corvidae’s data. You can think of accurate attribution provided by Corvidae as an accurate baseline for MMM testing to use.

If you are conducting MMM testing with ordinary pixel and cookie based digital analytics systems, then you are typically supplying conversion path & revenue data that is only 20% accurate, and you will also be supplying conflicting and cannibalistic data when combining AdTech data with site and app analytics, meaning the outcomes of the MMM can be misaligned to the true causes of conversion.

MMM is also a retrospective process that can take a long time and a lot of expense to complete.

They also don’t offer attribution insight to a useful granularity for digital marketing optimisation at the individual campaign level – let along individual adset, ad, impression, or keyphrase level, which is offered by Corvidae on a near-live basis.

Media Mix Modelling is a valuable and useful exercise, which is enhanced by the provision of accurate marketing data from Corvidae.

Do you utilise any third parties to build out a device graph?

No.

We trialled using device graph and other ID graph data in the early days of Corvidae’s development, but found that those data sets added little-to-no incrementality to the model’s accuracy.

Today, we use only first-party pixel, AdTech, and other client data, all compliantly collected.

What are your thoughts on total attribution/unified measurement?

Corvidae’s attribution model represents a unified measurement.

Marketing is rich in data and it if often said that there are many different attribution models for different data sets or measuring different aspects of marketing. I don’t share this view.

All marketing works towards a single goal – a conversion event of some kind. That might simply be dwelling on a particular page, or seeing a logo for more than three seconds, more than seven times in a defined time range, or a phone call or form submission and credit card transaction.

These are all conversion events, all measurable by Corvidae. Our model ultimately measures the influence of every event on some future conversion event, and does so to 95% plus accuracy. This represents a unified view of attribution, and a solid baseline for other, highly valuable insight to be derived using A/B tests, MMM, and other incrementality studies on top.

The power of Corvidae’s AI is its removal of all conflicting – cannibalistic – information, and a genuinely ‘golden record’ which can be held up as a ground truth for your marketing effectiveness.

So, there you have it. I hope the answers above have helped you to better understand the road to achieving cookieless attribution. For more information, you can download our eBook below. Or, book a demo of our cookie-free attribution software, Corvidae.