Google not allowing any form of identifiers : What does this really mean?

Last year, Google had announced the blocking of third-party cookies. Recently, Google has signalled that it shall not allow any form of identifiers across its suite of products. This article tries to deep-dive into what Google’s actions really mean in the context of evolving privacy landscape.

The evolving privacy landscape

Big tech companies are subject to more and more regulations around the world. The recent Information Technology (Guidelines for Intermediaries and Digital Media Ethics Code) Rules in India and the landmark News Media Bargaining Code in Australia are few examples of anti-trust laws that are coming up across the world.

Consumers are increasingly conscious of how their data is being used. A recent update of their privacy policy by Whatsapp has caused much furore. This has led to privacy laws being enacted by governments across the world. These privacy laws have mandated businesses to collect data in a manner which is compliant, and which protects the right to privacy of consumers. This evolving landscape is what forced Google and Apple to toughen their stance on privacy.

Following the announcement of the death of the third-party cookie, Google had come up with the concept of Privacy Sandbox.

In a press release earlier this year, they have evolved this into Federated Learning of Cohorts ( FLoC). They claim 95% accuracy as compared to cookie-based advertising.

Google also introduces FLEDGE, which involves an API connection for ad tech companies under a “bring your own server” model.

Google has very much been involved in the development of the IAB Transparency Consent Framework 2.0, and they recognize it across their partners, although they have kept their own inventory out of it. They have also come up with Additional Consent Mode- APIs like TCF for vendors who are not members of IAB. Now, Google has signalled that it shall not allow any form of identifiers across its suite of products.

Apple is one brand, that has taken an aggressive stance and upped the ante on trust. With the release of iOS 14, they have mandated privacy “nutrition labelling” on the App Store and mandated opt-in consent for identifier (IDFA) for tracking purposes.

What this means

2. Weakens the standing of alternative identifier based solutions: The use of non-cookie identifiers such as email addresses or phone numbers will no longer be supported by Google. That does not restrict the rest of the ad tech community from using it, but it potentially weakens their overall stance on privacy. Recently, India’s largest telecom operator Airtel has launched an ad tech platform providing access to its 320 mn customers. This has evinced a strong interest from advertisers. Telecom and email based data is known to have a higher data persistency than cookie based data.

3. The rise (and rise) of non-personal data: FLoC or Federated Learning of Cohorts is all about getting relevant insights from data without using any personal identifiers through aggregation and anonymization. In essence, it is like a segmentation process wherein we have deep insights into a cluster of respondents, but not individual information about any of the respondents. In a way, this is also data encryption.

4. Democratization of ad serving: The introduction of FLEDGE or First Locally-Executed Decision over Groups Experiment by Google points towards democratization of ad serving and make it economically viable for any advertiser to set up their own ad serving platform. This paves the way for a future where ad exchanges connecting DMPs and DSPs run on end smartphones of digital marketing managers.

5. Delivering privacy compliant performance: Marketers and agencies will need to deliver privacy compliant performance and not either-or. Thus far, it was a performance story with effective last mile attribution. Going forward, it will be a balancing act between privacy and performance given that Google FloC is untested and might not be as reliable as a cookie-based solution. Marketers and agencies cannot rely on insights offered by these walled gardens and might need to federated learning models of their own. The first step to this is effective harnessing of first party data along with a mix attribution model which considers the response delay (brand effect) of a campaign

A graduate from IIT Kharagpur and a gold medalist from IIFT Delhi, Ravi is a researcher at a legal think tank Enkrypt Council and an advisor to startups.