Google has opened up on its new approach to delivering and measuring ads sent to Chrome users once it finally rests the tired third-party cookie.
The incoming scheme has been the subject of much fear and speculation due to its market-shaking potential. Any feature implemented by Google will shake up the industry due to the popularity of its products - Google Chrome for example reportedly accounts for around 65.3% of web browsing activity.
The Drum explores what we can expect from FLoC (Federated Learning of Cohorts).
What is FloC?
The CMA is investigating the Privacy Sandbox and the ICO UK is again investigating real-time bidding impropriety after a Covid-19-shaped ceasefire. Internationally, the plates are shifting and regulation is inbound. Google needs to stay ahead of it.
The Federated Learning of Cohorts (FLoC) is a privacy-focused solution intent on delivering relevant ads “by clustering large groups of people with similar interests”. Accounts are anonymised, grouped into interests, and more importantly, user information is processed on-device rather than broadcast across the web.
Google is talking up the results from early tests simulating transactions “based on the principles defined in Chrome’s FLoC proposal”. It believes it has an “effective replacement signal” for third-party cookies. It will open for testing in March.
It says it is delivering 95% of the conversions per dollar spent when compared to cookie-based advertising. The effectiveness varies by the strength of the clustering algorithm and the strength of the targeted interest.
Google will make FLoC-based cohorts available for the public in March. By Q2, advertisers can start testing FLoC-based cohorts in Google Ads.
The Chrome 90 release in April will see the first controls for the Privacy Sandbox. For the user, it will be an initial on-off decision but further customisation is to come later.
Chetna Bindra, group product manager user trust and privacy at Google says: “We are more confident than ever that the Privacy Sandbox is the best path forward to improve privacy for web users while ensuring publishers can earn what they need to fund great content and advertisers can reach the right people for their products.”
The update is balancing the interests of the consumer, publishers and of course Google. There are many to appease.
Marketers are keen to be able to deploy their own audiences in Chrome. Criteo, NextRoll, Magnite and RTB House have brought forward proposals. It appears that a “trusted server” aproach could store information about a campaign’s bids and budgets – the scheme is called Fledge, formerly Turtledove. That’ll open for trials later this year.
Google’s inviting industry stakeholders to help shape how marketers will bid for these audiences.
When it comes to measurement, advertisers will “have to prioritize which conversions are most important for their reporting needs”. There will be noise added, and less data shared from devices.
The report concluded that efforts are being taken to protect people from fingerprinting techniques that allow covert tracking. Google talks up Gnatcatcher, its solution to make IP addresses, one of the easiest ways to track a web user.
A spokesperson for Marketers for an Open Web (MOW), a voice driving the CMA complaint in the UK, responded to the report. “Google’s proposals are bad for independent media owners, bad for independent advertising technology and bad for advertisers. The people who will be most significantly affected by this will be smaller local publishers and independent businesses – they will effectively be cut out of the open online advertising marketplace causing devastating damage to their businesses.
It said the claims of "collaboration and openness are disingenuous" with the proposals not being endorsed by the W3C. They concluded: "This is a monopolistic player attempting to consolidate their dominance by degrading the Open Web using privacy and collaboration as a veil of legitimacy."
And finally, the group asked Google to show its working on the claim the new approach was 95% effective as cookie solutions in simulatations. Below is a look at Google's methodology.
Update: This piece was updated Monday 25/01 17:30 to add MOW's response to the announcement. And then later to show how Google reached the 95% stat.