Expanding Privacy Sandbox testing
See updated guidance for the Relevance and measurement unified origin trial.
Today we shared an updated plan and timeline for Privacy Sandbox for the web and the path towards phasing out third-party cookies. Your feedback as web developers and site owners has been instrumental in demonstrating the importance of spending more time to get the proposals right and ensuring there's enough opportunity to test, integrate, and optimize the new solutions. This post includes more detail on testing plans, including our intent to increase the traffic volume in the unified Privacy Sandbox Relevance and Measurement origin trial in August, and extend the duration of the trial.
Since the start of the Privacy Sandbox initiative we've seen engagement from hundreds of developers, companies, and others across the W3C, industry events, our Chrome-hosted office hours, proposal repos, and more. In the past six months, we've also started sharing quarterly summary reports currently totalling at 145 aggregated feedback topics across 14 areas. Your feedback directly informs the design and development of the proposals and there are multiple routes open for you to provide it. Keep it coming!
The Privacy Sandbox project represents a broad and ambitious set of changes that aim to tackle cross-site tracking for the web as a whole. It proposes open standards everyone can implement rather than browser-specific features while ensuring that sites can use third-party services in a safe and private way. While phasing out third-party cookies represents a major milestone in overall project progress, that goal of addressing all forms of cross-site tracking is much wider! You should still expect the individual proposals and features to launch throughout that journey. It is likely that your site will be impacted in some way; you will want to ensure you understand how your sites and services are affected, and know which proposals and features you should be following.
Let's break down the current status and look at what you need to know to continue to test, provide feedback, and prepare for features on the path to launch.
Expanding the Privacy Sandbox Relevance and Measurement origin trial
The Privacy Sandbox Relevance and Measurement origin trial enables the ecosystem to run unified tests for technical stability and developer experience across Attribution Reporting, FLEDGE, Topics, fenced frames, and we will be adding Shared Storage soon. The trial is currently enabled for 50% of Chrome Beta users which has helped us to actively address early developer feedback and issues without disrupting users too much.
As the origin trial progresses, we want to give developers the opportunity to test the utility and effectiveness of the APIs with a meaningful proportion of real-world traffic. With the Chrome 104 Stable launch at the start of August, we will be expanding the trial to desktop users on Chrome Stable. We plan to extend the trial to mobile users starting from Chrome 105 Stable on Android. The origin trial is scheduled to conclude at the end of the 104 Stable period—we are requesting an extension through to Chrome 107 (late October) to enable further testing. This follows the standard practice of requesting origin trial extensions in three-milestone increments. We're committed to supporting testing through to launching the APIs for general availability.
You can follow the official request to extend the Intent to Experiment (I2E). We will also update the Privacy Sandbox documentation with implementation and testing guides.
If you deliver any of the services these APIs provide, then your participation and feedback in the origin trial is invaluable. As we move to larger scale testing, this is your opportunity to validate the proposals that meet your needs. There is no need for expertise in web standards or browser development—just your existing experience in your own field.
Once we reach a point where core functionality is sound and complete we plan to start shipping the APIs for general availability, likely in early to mid 2023. Over the course of the origin trial there is, by design, scope for the APIs to evolve based on testing and feedback. Individual features may launch while the overall origin trial is still ongoing. After launching we will continue to refine the APIs as we proceed through initial adoption and long-term testing.
Updating cookie behavior
The Cookies Having Independent Partitioned State (CHIPS) and First-Party Sets proposals provide a route to support cookies in cross-site contexts that do not involve tracking.
CHIPS
CHIPS allows developers to opt a cookie into "partitioned" storage, with a separate cookie jar per top-level site. Based on developer feedback during the current origin trial we made a number of fixes and improvements along with extending the trial period through to the end of Chrome Stable 104 at the end of August. Specifically, we have removed the more restrictive requirements for a __Host-
prefix and no Domain
attribute to enable easier migration for sites using cookies across subdomains, such as shop.example.com
and blog.example.com
.
With this positive feedback on both the proposal and the trial, we are hoping to ship CHIPS after the conclusion of the trial. As per the official process, you can follow the blink-dev mailing list for when we post the Intent to Ship (I2S) message.
That's an exciting milestone because for many use cases where you provide an embedded, self-contained service to another site such as a widget or API, this lets you get your updates done well in advance of the third-party cookie phase-out!
First-Party Sets
First-Party Sets provides a method to group affiliated sites intended to allow organizations that have multiple sites, such as different country-level domains, to still use their own cookies in these specific cross-site but first-party contexts.
Based on the feedback we received during discussion and testing of the feature, we're proposing a number of changes that aim to both address those issues and still satisfy the needs of the ecosystem. Specifically, we are proposing that sets be defined in terms of use-case-specific "subsets". We also propose that sites use the Storage Access API along with a potential extension to request cross-site cookie access. This replaces the proposal for the SameParty
attribute.
We will update the developer guide as work progresses. If you've already been experimenting with First-Party Sets or the use-case matches your needs, then this is a good time to follow the discussions and get involved.
Shipping user-agent reduction
We are currently reducing the information in Chrome's user-agent string. As of Chrome 101 in April, 2022, the minor or build version has been replaced with zeroes. The upcoming phases will also replace the OS/platform version and device model with fixed values. This will start for desktop from Chrome 107 in October, 2022 and for mobile from Chrome 110 in January, 2023. This timeline remains the same and is not affected by the changes to the phase-out schedule for third-party cookies with the fully reduced user-agent being rolled out by early 2023.
The changes to the string are intended to be backwards-compatible, so if you do not need those specific values then you will not be affected. However, if you do parse the user-agent string to extract the browser minor/build version, OS/platform version, or device model then you will need to migrate to User-Agent Client Hints.
Storage partitioning
Cookies are the most prominent feature used for cross-site tracking, but the Privacy Sandbox aims to tackle cross-site tracking as a whole—and that includes all forms of cross-site storage. In a similar way to how we already partitioned the HTTP cache in 2020, we also intend to partition storage APIs like IndexedDB and localStorage, communication APIs like BroadcastChannel and SharedWorker, and features that span both of those categories like ServiceWorker.
We have sent the Intent to Prototype (I2P) for this work which means we are progressing through designs and initial code for the various APIs. Within the current Chrome 105 Canary we are planning to have a flag available to enable local developer testing. You should expect these changes to progress through the standard Chrome development process as work completes which we expect to be in early 2023, in advance of the overall third-party cookie phase-out.
Developer documentation and support
To help you navigate Privacy Sandbox as a whole we have privacysandbox.com which provides the concepts, goals, and timelines for the project across web and Android. Here on developer.chrome.com/privacy-sandbox/ you can find the details for individual proposals, demos, testing and implementation guides, along with links out to wider resources for involvement.
We are holding regular developer Office Hours sessions across a variety of Privacy Sandbox topics. In each of these we bring in the engineering and product teams, run through a demo, and then answer your questions on implementation and testing. We publicize each session on the @ChromiumDev Twitter and on the mailing lists for the matching API. We're already providing a Japanese language session along with repeats for different timezones, but will also continue to improve the programme to post subtitled videos of the demos and making it easier for you to submit topics and questions in advance.
We have two more Attribution Reporting sessions coming up:
- English language - Thursday, August 11 @ 15:00-16:30 GMT
- Japanese language - Friday, August 19 @ 10:00-11:30 JST (01:00-02:30 GMT)
More details on the attribution-reporting-api-dev mailing list announcement.
We also have our developer support repo on GitHub. If you've run into an issue or have a question and don't know where to raise it, then post an issue there and we will help answer it or find the right place for you to get involved.
Providing and sharing feedback
While the Privacy Sandbox as a project was initiated by Google, the goal is that we are making proposals to change the web platform as a whole, not just feature changes in Chrome. That is an open and collaborative process across a large number of groups that includes browser vendors, site owners, and most importantly the people who use those sites and browsers—the users. While the resulting specifications are written in very explicit and formal language (since they need to define the process fully enough to implement) the process of making sure that specification does the right thing needs input from everyone.
We hear from a lot of companies who want to know who else is testing and how those results will be shared. It's up to you, as testers, to decide to make your testing plans and results public—and we strongly encourage you to do so! There are a number of public forums across the W3C, GitHub, mailing lists where you can share directly with other stakeholders. This might be as simple as stating that you're actively participating in an origin trial, whether or not you had all the material you needed to implement, or detailed analysis of the results of your tests. You can also publish to your own sites, blogs, or social accounts—especially where you have a specific audience you want to talk to.
Feedback highlights:
- Yahoo! Japan whitepaper on results from early Attribution Reporting testing.
- Google Ad Manager initial plans for Topics origin trial.
Our feedback page covers each of the different routes and the active ones for each API. You can also provide feedback directly to us via our feedback form.
In the end, by changing how cookies behave we're changing technology that has been part of the web for 28 years. The web belongs to all of us, and working through these changes to find that ideal mix that enables a more private environment while still enabling the rich, open ecosystem we all love will continue to require your input and direction. We're looking forward to the rest of the journey together.