Topics API: what changes can it bring? [FLoC comparison]
Topics API is the new take on privacy-focused and balanced interest-based advertising. The Topics vision has user-centric goals while building enough room for businesses to thrive. FLoC, the initial proposal for replacing third-party cookies, is gone due to scathing criticism. Let’s look at what Google learned after the first flop and whether Topics API is a suitable cookie alternative.
Recap of FLoC and third-party cookie death
The search for a brand-new approach towards advertising began after Google’s announcement to block third-party cookies. The journey of phasing out cookies and introducing a worthy candidate proved strenuous.
Thus, Google delayed the 2022 deadline until 2023. Depending on the success of Topics API, the date could keep changing.
The new Topics model replaces FLoC (Federated Learning of Cohorts). The latter prompted user division into groups of people with similar interests. However, the lack of interest from publishers and browsers’ refusals to adopt it pushed Google back to the drawing board.
What is Topics API?
Topics API is one of the Privacy Sandbox proposals to supply a feasible and more private personalized advertising. Topics operates by figuring out interests from your activities and using them as the basis for online ads. The interests rotate each week, suggesting fairly quick ad-tailoring.
As Google explained, browsers indicate topics you have shown interest in throughout a week. It means looking at your activities and figuring out which topics you engage with most frequently. For instance, a website offering beauty products could have a topic of “Bath & Body Products.”
So, Topics API does not need to resort to tracking the sites a user visits. Instead, it will only obtain the final interests, with a 5% chance of one being randomly selected. When users visit websites displaying ads, their ad tech platforms generate ads according to details supplied by the Topics.
The aim is to escape the need for revealing specific sites users visit to external parties, including Google. It is how third-party cookies functioned, allowing websites to track consumers’ activities across the web.
Taxonomy: interests in the Topics API
Topics API would operate with a taxonomy, listing all possible interests like fitness, poetry, or design. Initially, Google would curate the items, but it expects trusted external contributors to help maintain them.
One of the priorities for this list is its size. It has to be small enough to ensure that each topic would get a decent amount of browsers. Furthermore, Google mentions the need for careful administration to avoid sensitive interests in its Topics API. They should not relate to ethnicity or sexual orientation.
Currently, Google has announced around 350 potential topics. However, the list could go as far as reaching a few thousand entries.
Here is a snippet of possible topics:
Quick rundown on how Topics API facilitates ad delivery
Topics API can seem confusing, especially due to the incomplete information on its operation. Google has not set a date for trials, but tests should generate more insights into this cookie alternative.
For now, we can divide the personal ad delivery via Topics API into these steps:
- The browser learns your interests from your habits (visited websites). It retains information for three weeks. After that, it generates new interests from the Topics.
- Each website represents a specific interest like workout tips can end up in the fitness category. Depending on the frequency of your visits, the topic might become applicable to you.
- If sites do not have categories, machine learning algorithms predict the topic based on the domain name.
- Once Topics API has listed your interests, it needs users to visit websites supporting this system.
- Websites with Topics API support will receive three of your interests generated throughout the three-week period.
- Sites’ advertising partners will receive your interests and decide which ads are suitable.
- Finally, you will see personalized advertisements without having your activities tracked across the web.
So, how is Topics API different from FLoC?
Topics API is the system Google developed after FLoC. It claims to have considered all feedback and criticism for making Topics better. For now, it seems promising, but more information should become available after testing begins.
Criteria | Topics API | FLoC | |
---|---|---|---|
1. | Main idea | Browsers collect top interests from three weeks based on users’ browsing history. Ad platforms receive three topics and select ads based on them. | Assigns users to groups (cohorts) of people with similar interests. Anonymous cohorts would contain thousands of users. |
2. | Privacy perspective | A more private option for delivering ads without exposing the consumer. Still shares consumers’ interests. | Similar problems as with the third-party cookies. Still possible to distinguish individual users. |
3. | Opt-out options | Possible. | Possible. |
4. | Efficiency | 300-2,000 possible topics. Might be too generalized for relevant ad targeting. | Over 30,000 cohorts. Possible to supply more targeted advertisements. |
5. | User control of ads | Users can review and remove the generated topics. | Users can review and remove themselves from cohorts. |
6. | The main problem | Still less private than contextual targeting. Could seem impractical for advertisers. Needs more testing. | Possibility to identify users’ via fingerprinting and lack of transparency with data interpretation and cohort assignment. |
7. | Sensitive topics | Promises not to generate any sensitive categories (like gender and race). | Promises not to generate any sensitive categories (like gender and race). |
The comparison relies on the information currently available about Topics API. The features and functionality might gradually change.
Is Topics API a decent idea?
Topics API has the potential to be a more privacy-preserving option than third-party cookies and FLoC. However, it still prioritizes interest-based advertising, meaning that consumers’ data gets used for advertising purposes. Luckily, there seems to be less room for exploitation and more insights into how users’ activities get interpreted. For instance, experts emphasized the uncertainty of how third parties could exploit FLoC IDs.
Still, Topics API is in its early stages of development. The idea needs polishing, and tests should reveal its efficiency and immunity to possible misuses. Whether other browser vendors will express interest in implementing Topics is also a question. Considering the refusal of FLoC, the same scenario could happen to this proposal.
Another important note is that Topics API could be just an addition to the techniques equipped for delivering ads. There is a lot of augmented data on users, like their current actions or contextual cues. So, the new mechanism could become another signal for advertisers to use in correlation with other tactics.