Back to Blogs
By Aqilliz
Published on July 23, 2020
As we gradually reach the end of July, we’re now only roughly 18 months away until Google’s Chrome browser will phase out support for third-party cookies for good.
Can we really say we’re ready?
In the first half of 2020 alone, advertisers are still largely relying on tried and tested solutions in identity management which allow advertisers to identify users as they traverse the internet—this is vital to the online advertising industry, as publishers look to display the most relevant ads possible to their visitors.
And yet, these regulatory headwinds simply aren’t new. As a publishing executive stated at this February’s Interactive Advertising Bureau’s (IAB) Annual Leadership Meeting, “How long have we seen the regulation freight train coming? Twelve years? Now it’s here, and all behaviour has to change”.
To mitigate the incoming demise of third-party cookies, advertisers and publishers have been pulled in by the lure of walled gardens. Admittedly, this is in part due to the swathes of data—such as login data matched to actual individual profiles—that they have at their disposal. Deterministic rather than inferred, this naturally offers greater advantages for identity resolution as well as accurate targeting.
So, where do we go from here as we move away from third-party cookies? What are some of the options we have ahead of us?
Setting a Standard for Experimentation
Experimentation is essential, but it’s important that industry players also look at the sustainability of these offerings. As players continue to innovate in a bid to weed out the most promising solution, the industry might be at risk of forgetting that it’s not only a matter of developing something that works, but something that can be adopted across the board.
  • Do they provide common naming conventions?
  • Do they offer common standards?
  • Do they champion transparency?
  • To what extent do these solutions allow us to depart from centralised, data silo models?
These are the questions we should be asking ourselves, as we evaluate propositions for “better infrastructures” that are agile enough to adapt over time. We could have a plethora of identity solutions, each able to deliver a degree of success, but not all publishers would be able to take them on—they simply wouldn’t be able to.
In light of that, the industry needs to examine a strategy that places precedence on efficiency as much as compliance, across a fragmented advertising ecosystem.
Taking on a Shared (ID) Approach
Today, shared identifiers (shared IDs) look to create a common standard in how user data is identified and shared across publishers and AdTech providers. This allows the matching of cookies to take place in mere seconds. To date, a myriad of Audience ID solutions have been proposed by various leading AdTech players.
While each solution proposes different benefits, they generally offer a shared, deterministic customer identity based on first-party publisher data for advertising purposes and some still continue to leverage cookies in some shape or form. This customer identity is commonly arrived at via single sign-up partnership with individual publishers or a common sign-on across a publisher consortium. At present, however, even solutions such as these are grappling with their incoming redundancy, with some having been blocked by browsers such as Mozilla’s Firefox.
And until a suitable alternative is found, it’s likely that industry players will continue to use existing offerings until the very end. IAB Tech Lab SVP Jordan Mitchell argued that eliminating the shared ID system “before the very last third-party cookie is served… would lead to missed commercial opportunities for both advertisers and publishers” as they struggle to identify users across the online landscape which they traverse with different devices.
To grapple with this, it’s clear that while such efforts will likely serve as the phase of experimentation that we’re likely to see in the months to come, the industry needs to look beyond solutions that are built on the back of cookies as a whole.
Playing the First Party Game
Meanwhile, other alternatives are looking to uphold new identifiers altogether, built out of first-party data points such as email addresses and phone numbers. This is a promising sign, as a transition to first-party data is certainly a move in the right direction: it represents a movement towards data that is not only ethically acquired, i.e. with user consent, but data that is more likely to be of greater quality.
As 2022 approaches, it’s likely we’ll continue to see new identifiers emerge as part of a period of trial and error. However, this will invariably lead to a crowded complex, programmatic supply chain, as ad buyers and sellers look to match up different identifiers, each with their own specific uses.
Publishers will also need to invest in developing new infrastructures—an entry point, so to speak—to facilitate the collection of first-party data while obtaining tacit consent from users directly. Sure enough, this will be a paradigm shift for publishers and users alike who, for decades, have been accustomed to a streamlined journey of interacting with websites across different devices.
Over time, marketers will also need to make investments in scalable customer data integration systems, to merge identifiers, de-duplicate, as well as remove them when requested by users.
Collaborative and Decentralised
In taking the best of what these two existing offerings can bring, we can look at a system that both encourages collaboration while presenting a system of decentralisation. This ensures that data is secured and treated in such a way that is private by design, and therefore compliant with data privacy frameworks. Simultaneously, data silos are unable to proliferate, encouraging publishers and brands to share their data points within a federation.
This is the cornerstone of our approach at Aqilliz, which aggregates data from multiple sources into a virtual database. Before data points are shared, each one is appropriately encrypted with cryptographic algorithms such as differential privacy to ensure that they cannot be reverse-engineered to reveal the data in its raw state. By applying additional measures such as federated learning, data analysis takes place across a network of data sources in a decentralised manner—rather than everything being analysed in a centralised location, machine learning models are deployed to these sources which ensure that data never leaves local storage. Out of this, interest graphs and audience profiles are produced on an aggregated basis, stored in a distributed manner for utmost security.
Other models, such as those proposed in Google’s Privacy Sandbox, utilise a similar federated approach, grouping user data derived from browsers as “flocks”. This enables a form of interest-based ad targeting through audience group identifiers that are borne out of the Chrome browser itself.
As we look to 2022, there’s certainly a long way to go as the industry arrives at an identity management solution that can satisfy browser operators, regulators, and discerning consumers alike. If there’s anything for certain, it’s that the existing models that we have at our disposal are simply not enough.
A privacy-centric model of the internet is well on its way and we can no longer afford to build on the very foundations that have led to this tipping point in the first place. And while it may be a tough (and costly decision) to make, we simply had it coming.
We’d say it’s finally time for the industry to look forward and adapt.
Last month, we published our very first Insights Paper, which looked to reimagine the future of identity management with federated learning and distributed ledger technology—head over to our Resource Library to give it a read, if you haven’t already!