The future of attribution is cookie-less, privacy-secure, enables cross-media measurement and is highly effective

What is attribution? The (Google) dictionary definition is: “The action of regarding something as being caused by a person or thing.”

So, in our advertising world, it’s the essential practice of being able to demonstrate that doing (or not doing) something in a campaign leads to a measurable, hopefully positive, effect – such as increases in sales, intent to buy, store visits and so on. With an accurate understanding of what is working, brands and agencies can reap the rewards of better campaign ROI, and contributing publishers and platforms will get the credit and budgets they deserve. The future of attribution is the engine room of a truly effective, and potentially more transparent, digital advertising ecosystem.

Attribution is not a new idea. In the recent past, attribution in digital media relied heavily and almost exclusively on the third-party cookie, and the last click wins model. While this served some parties well it didn’t paint a picture of the whole path to purchase, across multiple channels.

It is well documented that major changes are unfolding, which will further challenge cookie-based attribution in digital (Google third-party cookie deprecation, GDPR, etc.). So we need to reach the future of attribution as soon as possible – and it’s nothing to fear, because it will be far more comprehensive than what is relied upon now, and will be able to cover a lot more of the contributing channels than just digital media.

So, what is the future of attribution?

Simply put, it is about having quality samples on which to run careful experiments; samples that are representative of the total population. It is about measuring, understanding and predicting conversion probability at scale.

Existing digital media infrastructure can be fused with tried and tested research techniques to enable extremely accurate predictions across digital and other campaign channels, at scale.

AudienceProject’s Director of Research, Geert Laier Christensen, explains more: “It’s important to understand the basics. The attribution goal is to estimate the overall campaign effect on the campaign KPIs – sales, store/website visits, brand lift, etc.

“The simple logic behind any effect measurement (and not just in advertising) is to ask the question: what would have happened had we not run the campaign. We can estimate that using historical data in econometric models, but this is a very resource-consuming task and it will often lack precision because the particular combination of ‘causes’ influencing results is unique to the period where the campaign has been running (e.g. at the time of writing, you would, obviously, not be wise to use the ‘Corona Quarter’ as the baseline for normality).

“The use of ‘the experimental’ method, where control geo-zones (exposed and non-exposed) are carefully selected and used, gives a powerful predictor that overcomes these problems. The figure below illustrates the logic. Conversions (or the relevant ‘lift’ KPI) are tracked geographically before and during the campaign. Then a statistical model is used to predict what the conversions would have been in the campaign period, had the campaign not been running. The effect is the difference between the actual and the predicted amount of conversions.

“But what about the practicalities? For this to work, we need to be able to do two things: we need to be able to exclude some geo-zones, and we need to be able to accurately track impressions and conversions geographically.

“This is where the technology comes in – to keep the experiment pure, enable relevant geo-selection and adjust for movement between zones (e.g. commuters). The technology can also add clarity around demographics, frequency and so on where panel-based audience tracking is deployed in tandem – and device graph technology will help to add insight into the true frequency of individuals, as well as controlling for geo-zone ‘spillage’.”

The beauty of the attribution future is that you don’t even need to track the individuals who come into contact with the campaign for it to work at a fundamental level – a privacy campaigner’s dream. And, multiple channels can be understood at the same time – i.e. run linear TV alongside digital and look at the results. Or understand the impact of walled gardens and other channels in combination – where it is currently not possible to understand users moving between them.

So, to summarise, we see the future of attribution as a combination of proven research practices, enhanced by the careful use of technology. It is cookie-less, privacy-secure, enables cross-media measurement, and is highly effective (as recent case studies show). And the best bit – it is actually all possible today.

Find the full IAB UK Guide to Digital Innovation 2020 here.