Just a few weeks ago we talked about cookies and tried to take stock of the revolution in the field of identification tracking for marketing purposes, with the progressive abandonment of third-party trackers as a signal of attention to the privacy protection and security of user navigation. These two points are at the center of the new technology announced by Google, which is called FLoC and is a first appropriate alternative to cookies.
What is Google’s FLoC technology
Federated Learning of Cohorts, or FLoC, is an advertising technology that basically groups “large groups of people with similar interests”, using “data collected in the browser, not stored anywhere”, which analyzes to create an “anonymous individual hiding in the crowd”, as written in a long article by Chetna Bindra, Group Product Manager, User Trust and Privacy.
The API will be integrated as a browser extension within Google Chrome and intent on offering “businesses a new way to reach consumers with relevant content and advertising through large groupings of people with similar interests (cohorts)”, but with the feature of processing information at a device level (smartphone, tablet, laptop) “so as to keep private on the browser the web history of each person”.
How works targeting with FLoC
The data shared and used to target ads comes from the broader cohort – the “crowd” of thousands of people, but anonymously compared to previous technologies.
Specifically, the software uses machine learning algorithms to analyze user data and then create a group of thousands of people (the cohort) based on sites visited by a single individual. The data collected locally by the browser is never shared, while it is the data of a much larger cohort of thousands of people to be shared and then used to target ads.
Tested, the targeting system works and gives effective results: after “Simulations based on the principles defined by Chrome for Floc, Google’s advertising teams have compared this strictly privacy-oriented solution with third-party cookies”, revealing that compared to the generation of “interest-based audience segments, Floc provides a signal that can effectively replace third-party cookies“.
Great results on conversion tests
In concrete terms, advertisers can expect to “see on average at least 95% of conversions per dollar spent compared to cookie-based approaches”, according to the results of tests on FLoC to reach in-market and affinity audience segments (Google Audiences).
An impressive figure, which “depends on the strength of the algorithm used by Floc to run clusters and the type of audience segment that you intend to achieve”, and which proposes the Federated Learning of Cohorts solution as an interesting opportunity for users, publisher and advertisers.
The technology will be made available on Chrome as early as March, with origin trials that will allow you to use FLoC-based clusters for public testing, and forecasts in the second quarter of 2021 will be extended to Google Ads. Meanwhile, you can begin to familiarize yourself with the system by testing some simulations following the principles outlined in a whitepaper by FLoC published on Github (where Google has also uploaded the code base for FLoC open to all).
A way to overcome cookies
Sara Fischer on Axios briefly reconstructed what is happening in the media landscape and listings following controversies and criticism around third-party data (which includes many types of cookies and user data collected indirectly through browsers or websites), especially due to the scandals that have shown how they are often purchased and sold on a large scale through online data exchanges.
And so, the major browsers have decided to (or have been forced to) gradually give up cookies, for decades the main tool through which most advertisers turned to online users, because privacy concerns have made this path less practicable, also in the light of the legislative showdown on online data privacy in the United States and Europe.
Finding an alternative to cookies is a huge challenge, given that the entire digital advertising ecosystem, worth 330 billion dollars globally, has been built around them.
It is not surprising, then, that Google is making a major privacy-oriented effort on the one hand and, on the other hand, to still allow tracking for marketing purposes, with a more gradual and collaborative approach to its privacy changes than Apple, which has been criticized – mainly by Facebook – for introducing radical changes to its user tracking function “Identifier for Advertisers” (IDFA) without providing advertisers with a better alternative.
As Chetna Bindra explained to Axios herself, for Google “an innovative approach in the industry made more sense” and the company focused on “not immediately block third-party cookies”, but in improving them thanks to technical innovation and collective contribution: “We want to involve the entire advertising community and we really rely on the kind of collaboration that is essential to bring about such a huge change,” she explained.
Privacy Sandbox and other Google’s projects
The news that Google is close to proposing a technology that will replace cookies has clearly created great expectations and interest, because one of the main problems of the gradual elimination of cookies from targeting Internet ads is precisely the absence of a viable alternative solution.
The effort of Mountain View is part of a wider initiative, launched on Chrome as early as 2019 and called “Privacy Sandbox“: a set of rules open to the entire advertising industry to gradually eliminate cookies in a privacy-friendly manner. In addition to this, the company stated that it has other proposals in place to replace cookies, so it is not guaranteed that Floc is the final answer (although the results obtained so far have been very encouraging).
The work of the Californian company started from the consideration that “advertising is essential to keep the web open and accessible to all, but this ecosystem is at risk if privacy practices do not fit people’s expectations”who want to be sure “that their identity and information are protected when they browse online”.
The new systems to create an audience
Among the various features of Sandbox Privacy there are also proposals to help marketing experts to create and manage their own audience, obviously without the use of third-party cookies, for example with remarketing to reach those who have already visited a site.
The proposal published on Chrome is called FLEDGE and has been fine-tuned with the support of several members of the ad tech community, including Criteo, Nextroll, Magnite and RTB House: thanks to the use of a “trusted server” (defined in compliance with certain rules and principles), the system is developed specifically to store information on the offers and budgets of a campaign.
Proposals such as FLoC and FLEDGE explore privacy-friendly alternatives to reach relevant audiences, while other aspects of Google’s work have focused on improving advertising auctions and helping those who buy advertising space decide how much to offer to show an ad to a certain audience segment.
Measuring conversions with Privacy Sandbox
The Google project also contains a number of technologies useful to measure the performance of a campaign without third-party cookies, respecting consumer privacy and, at the same time, the main needs of advertisers, such as reports on user interactions with content (which allow auction bidding models to identify occurrences within the data) and aggregated reports (which provide accurate measurements on a group of users).
APIs use various techniques to provide conversion reports that respect user privacy, such as aggregating information, adding noise, and a limit on the amount of data sent from a device.
Systems to prevent advertising scams and block fingerprinting
The technology has been extended to try to solve two other common (and serious) problems, namely advertising fraud and fingerprinting.
To help companies distinguish real visitors from fraudulent traffic – a crucial factor in advertising-supported web health – since last July Chrome has been testing the Trust Token API, that is precisely to identify genuine traffic without endangering the identity of people. Moreover, in the coming weeks Chrome will support a new feature for Trust Token that should improve the detection of fraud on mobile devices, always in the protection of privacy.
It is instead called Gnatcatcher the Chrome proposal to mask an IP address and thus protect the identity of a person without interfering with the normal activities of a website, in response to illicit fingerprinting techniques and those invisible or opaque practices that share data about individual users and then trace them secretly (for example, using the IP address of a device to attempt to identify a person without his knowledge or without the ability to disable tracking).
The future of privacy on the web
Google has therefore made its move to improve the privacy of online users and at the same time ensure the work of publishers and advertisers, which defines “element necessary to finance quality content and to reach the right people with relevant information”.
FLoC’s first results, the advancements in the development of APIs and the encouraging dialogue with the industry lead us to think that Privacy Sandbox can impose itself as a new reference for the operation of advertising and measurement products on the web.