Find out why marketeers can no longer rely on third party targeting and attribution, and what they’re using instead.
It’s official: third party targeting and attribution is now dead in Safari. For regions where the iPhone dominates, marketeers are flying blind in the fastest growing area of advertising revenue and customer engagement – the mobile.
Even then, given that Chrome has 50-60% share of the total browser market across all devices, the second (Safari) and third (Firefox) largest browsers don’t allow third party cookies at all.
As a result, marketeers have been looking for an alternative option – with varying success.
Option A: Moving from third-party cookies into a first party world
With third party cookies, previously the lifeblood of targeting and attribution, now dying, some in the industry have started to move into the first party world.
Facebook, Alphabet and Microsoft all released a first party tracker soon after Apple’s initial release of ITP. A first party cookie can be defined as a cookie that is relevant to a specific domain. First party cookies are used for many things on websites such as remembering a basket, where you might have visited on the website and might want to visit again, passing information through a checkout process and much more.
The different treatment of first party and third party cookies is not insignificant. Both browsers and legislation recognise some instances of this type of cookie to be more essential in the effective operation of a website. As a result, they don’t immediately block or delete it. Even Safari, with the latest release of ITP 2.1 allowed them to exist on the browser for a period of seven days (that is a rolling seven days from its last use).
But, while first party cookies might be helpful for tracking activity on a single domain, they become less useful when trying to attribute customer engagements to conversions across multiple touchpoints on multiple domains.
Without the different activities being linked to a specific browser holistically, as the ‘golden bullet’ for attribution it is limited as it really only can work for one channel (i.e. Facebook) and one domain (i.e. a brand using Facebook). Combine this challenge with Apple’s ITP 2.2, which now shortens first party cookie lifetime down to a rolling 24 hours, in itself it cannot support proper lifecycle attribution.
In any case, seven days (soon to be 24 hours) to complete an end-to-end customer engagement within a lifecycle is normally not long enough. So depending on this as a single solution for attribution across multiple touchpoints will never give you a complete picture.
Option B: Moving from Deterministic attribution to Probabilistic attribution
Deterministic attribution is matching to 100% certainty of an engagement to an outcome. An online identifier residing in a cookie might help you do this.
Probabilistic attribution is having a number of pieces of information and putting them all together, getting to a statistically significant probability that that this then person to which an engagement probably relates, and has led to the eventual outcome.
In the absence of matching pseudonymous IDs in a deterministic way, with the advent of more sophisticated machine learning capabilities and to a lesser extent AI in the current environment (maybe more in the future), it has become easier to read and interpret signals that could imply statistically significant links between several events or scenarios.
With the average number of internet connected devices last year reaching an all-time high at 3.5 per household, much has been done in the use of probabilistic matching in attempt to understand cross-device engagements by single customers across their multitude of devices.
Historical data is critical in the process to develop machine learning, and for AI to identify trends and scenarios that can be linked together. The main challenge around this is volume. To be able to set probabilistic matching off in an algorithm, there needs to be a lot of data as a basis.
Combine this with data points in probable matching never being concrete, and the exponentially different ways in which a browser might interact with each different brand, the variations and predicting variations becomes challenging.
Much of probabilistic matching for attribution is based on leveraging multiple datapoints in the different browsers to triangulate to a specific browser. Again, a walled garden has intervened (Alphabet this time) to say that they are going to make it difficult by ‘frowning on this practice’.
Why it’s time for Option X
With pitfalls in the two different approaches being adopted, there has be another way. This is Option X – the Hybrid Attribution approach.
To learn how Hybrid Attribution works, and why an increasing number of brands are now using it, download our free white paper.