Blog post

Demystifying Google Privacy Sandbox

No items found.
Key visual for Demystifying Google Privacy Sandbox
Written by
Stefan Deml
Published on
March 19, 2024

What Privacy Sandbox can and can’t do for digital ad targeting in the post-cookie world

Recommended reading

Post-cookie prospecting playbook

Our playbook breaks down the current options available to brands for targeting audiences on the open web — and how they stack up when it comes to reaching net new customers.

Key visual for post-cookie prospecting playbook

Google is making good on its promise to disable cookies by the end of this year. And to avoid leaving marketers in the lurch, it has introduced the Privacy Sandbox, consisting of several new APIs and functionalities added to the Chrome browser. However, as confirmed by the latest IAB Tech Lab report on the subject, these new solutions don’t cover every use case — meaning if a brand relies solely on the Privacy Sandbox, it may be missing valuable chances to connect with relevant audiences.

In this article, we’ll break down where the new APIs can help with targeting, and where brands may need some additional tools to achieve the maximum balance of precision, reach, and privacy.

Google has divided the key proposals which make up the Privacy Sandbox into five categories:

  1. Strengthen cross-site privacy boundaries
  2. Show relevant content and ads
  3. Measure digital ads
  4. Prevent covert tracking
  5. Fight spam and fraud on the web

While the Privacy Sandbox covers a range of privacy and ad operations challenges, the most pressing issue that brands are facing as a result of third-party cookie deprecation is how to reach their best audiences. So for this article, we’ll narrow the discussion down to category 2 from the above list (“show relevant content and ads”) and the two APIs that aim to pick up where cookies left off: the Topics API and Protected Audience API.

Topics API

The Topics API allows marketers to tap into users' interests for targeting without delving into their browsing habits. It enables the browser to monitor and log topics of interest to the user, as inferred from their web browsing behavior. This data is stored locally on the user's device. The API allows marketers access to these topics without disclosing further details about the user's web activity.

However, this approach comes with limitations, the first of which is reach. In Europe, users will have to actively opt in to interest-based data collection with Topics, greatly reducing the likelihood that it will be used. Additionally, the capability of Topics to build interest profiles is limited to a maximum of five interests per user. This makes segmentation and modeling with Topics less attractive.

Protected Audience API

With the Protected Audience API, a user's browser retains advertiser-specified interest groups linked to it and conducts on-device auctions to display advertisements. These clusters are established based on user behavior across sites (similar to cookies), but won’t be shared with third parties (unlike cookies).

This approach excels in traditional retargeting scenarios for users who have previously interacted with a brand or website but did not complete a conversion action — like making a purchase or signing up for a newsletter. However, it falls short for brands seeking to reach entirely new audience segments consisting of people with no prior interaction with the brand.

So how can brands prospect with the Privacy Sandbox?

As it stands, they can’t, as both exclusion targeting and lookalike modeling/targeting aren’t supported by the new APIs.

While both methods are useful for prospecting, lookalikes are more sophisticated than exclusion targeting.  Exclusion targeting systematically filters out existing customers to focus solely on new customer acquisition. But it does so without differentiation, targeting all non-customers without regard to their likelihood of engagement. In contrast, lookalike modeling adopts a more refined approach by not only excluding current customers but also ranking potential new customers based on their similarity to existing ones. This method allows for a more targeted, efficient, and sophisticated search for new customers.

This means if brands want to reach net-new customers, they’ll need to look outside of the Privacy Sandbox solutions we’ve outlined above. And brands essentially have two options for lookalikes moving forward: lookalikes in walled gardens or a data clean room with lookalike audience creation capabilities.

Lookalikes within walled gardens

Many marketers will already be quite familiar with the lookalike audiences features available within the majority of walled garden platforms. And while this is certainly a good option to support brands’ prospecting efforts, they’ll still be missing out on the opportunity to connect with users on the open web. Users spend 34% of their online time within walled gardens, as opposed to 66% of this time on the open web.

Data clean rooms with lookalike modeling capabilities

A great solution to ensure brands can continue prospecting on the open web is to use a data clean room solution that enables them to generate a lookalike audience. Methods vary, but the basic idea behind these solutions is that brands can upload their data into a data clean room with that of their chosen publisher. Then, the clean room technology can find new audience members within the publisher audience who have similar attributes to their existing audience.

But while these solutions fill a lot of gaps left by the Privacy Sandbox as well as walled garden solutions — enabling brands to continue prospecting on the open web — not all of them offer adequate privacy guarantees. In fact, oftentimes a brand will still be required to share portions of their first-party data with the publisher, such as any users that match with the publisher’s. In many setups, the publisher will use this data to then build the lookalike itself (necessitating access to brand data).

Not only does this bring about extensive (and expensive) legal arrangements, but brands also lack control over the treatment of their original audience. The only option that allows brands to have the reach and precision they’re used to, while still ensuring their data remains private and can’t be seen or accessed by any other party they collaborate with? Privacy-preserving data clean rooms.

Privacy-preserving data clean rooms with Decentriq

Decentriq's Lookalike Clean Rooms ensure data remains encrypted and inaccessible to any collaborating entities at all times, thanks to our implementation of confidential computing within the solution.

It’s tailored for business users, so there's no need to have data scientists on brands’ teams. And they can immediately begin uncovering insights by inviting publishers they directly collaborate with to the clean room or choosing from our network of publisher partners.

One additional note: Keep in mind that Topics API as well as Protected Audience API are restricted to users of the Chrome browser (used by only 50-60% of European users). By using privacy-preserving data clean rooms, brands can benefit from reaching all users of the open internet while still meeting privacy requirements.

Where can my brand find out more about cookieless prospecting?

Check out our “Post-cookie prospecting playbook” that breaks down the current options available to brands for targeting audiences — and how they stack up when it comes to reaching new customers.


Recommended reading

Post-cookie prospecting playbook

Our playbook breaks down the current options available to brands for targeting audiences on the open web — and how they stack up when it comes to reaching net new customers.

Key visual for post-cookie prospecting playbook

Related content

Subscribe to Decentriq

Stay connected with Decentriq. Receive email notifications about industry news and product updates.