Since the early beginning, Neodata Group is testing TOPICS API, one of the Google Privacy Sandbox Proposal related to interest-based advertising.
The Privacy Sandbox initiative, launched in 2019, “wants to create technologies that both protect people’s privacy online and give companies and developers tools to build thriving digital businesses”. It does so, by following the approach of the definition of web standard, including public discussion, testing and release. Being part of the ad tech ecosystem since its foundation, Neodata has decided to join the public discussion and provide feedbacks, aiming at keeping providing services and products to its customers – advertisers, publishers, and media agencies – in order to interoperate in a continuously changing market.
The Italian company is now used to be the only representative of its country within the Origin Trials. In fact, Neodata’s Team had already tested the Floc API solution, which then has been stopped by Chrome itself after taking into account feedbacks and test results. The team then offered to be the tester also for the Protected Audience Solution, a proposal addressing the remarketing use case.
Topics API, then, has its pros and cons, as Salvo Nicotra, Neodata Group’s CTO, has shown during a webinar on May 3rd. Let us go through it.
A three-years old story
Back in 2020, Neodata Group – together with UPA and Google itself – hosted a first webinar on this… topic. In January of that same year, Google Chrome had announced the Third-Party Cookie deprecation, and the company – as well as many others around the world – tried to imagine a possible tech scenario out of its since-then comfort zone: what could be the perspectives for a Cookieless data collection?
Well, since then the market has come a long way – it also went through a global pandemic. After the announcement, several solutions were proposed, and some working groups have tried to reason on them. Some of them have proved valuable to be tested, but, until now, none has proved to be THE solution to the Cookieless problem. Therefore, in 2021, Chrome postponed the cookie deadline to the end of 2024, to give the market time to find an efficient way out.
2022 saw the Floc API testing, which though proved inconclusive, and the solution itself was discarded. Here, you can also read Neodata Group’s considerations about Floc, as the company participated in the “Origin Trials” from the very beginning.
In the meanwhile, another two solutions were launched: Topics API and Protected Audience (formerly FLEDGE). From now on, testers have a little more than one year to try these solutions and provide feedback on their efficiency. Here you will read some of Neodata’s considerations about Topics API.
How does it work?
How does Topics API work? How does it provide chances to successfully address profiled customers and prospects?
From a technical point of view, Topics API is an on-edge-intelligent application that is able to compute up to 5 topics per user and provide ad tech providers with up to three interests. The execution of the procedure happens in the Chrome browser using a classifier that transforms the hostname of a visited page into categories belonging to a specific taxonomy. In this way, without using any identifier associated with the user, it is possible to deliver an ad that matches the calculated interests, i.e. they are more relevant for the user. If you wish to know more, please refer to the Chrome website for a more detailed description of the algorithm.
In this phase, Privacy Sandbox relies on the Chrome Origin Trials’ methodology to allow developers to test the functionalities with real data upon a small percentage of traffic (5%); Neodata Group, leveraging on the existing integration in clients owned media, is able to activate this test without any effort from the publisher side. However, as Google lately announced, they have the intent to extend the percentage of traffic to 100%, later in 2023.
Neodata has decided to propose its customers to participate in the test phase, explaining the methodology and the impacts of the test in the market. Most of its customers, and counting, have accepted to join with enthusiasm and high expectations, thus demonstrating that the market is willing to find solutions and be able to continue to communicate with audiences in this new scenario.
The first results
It is time to stop, take a step back and watch what we have done. Considering the testing period of two months, March and April 2023, Neodata Group’s Team can show a few numbers to begin with:
- The percentage of traffic enabled to the Topics API stays below the 5% limit, defined in the Origin Trials.
- All the 350 topics defined in the Topics Taxonomy have been returned during the test (100% topics coverage).
- 1.3 is the average number of epochs for each user, i.e., only 30% of users returned after one week in the period of observation.
- 11% are the users who changed profile over different epochs.
These numbers are highly preliminary and based on 1k signals; they are not statistically significant, of course. However, it is possible to make some considerations:
- The definition of the epoch duration (currently one week) may differ according to the categories of websites. For example, brands and e-commerce websites need shorter sessions due to lower return of visits.
- For a vertical website, Topics API can picture an “outside view”, which is able to provide some useful insights about their own audiences.
- Taxonomy based only on hostname leads to small variance of interests; a taxonomy based also on the entire URL would increase this value.
In the next phases of the experimentation, considering an increase in traffic and participation, it would be possible to better understand if these trends will be maintained.
Possible applications
At this point, one should ask himself: what will we be able to do with Topics?
The standard application is to “show advertising (or other kinds of, ed.) contents relevant for the user”. This is the definition that Google gives on its website.
But there are several possible scenarios that Salvo and his team envisioned, while testing the solution. “First of all, Topics signals could be used on the edge, together with other contextual signals and interest groups (eg. Protected Audience)”, Salvo said. “Then, First-Party Data and the Global Audience Analysis could be enriched with information coming from Topics, which, collected over time, could be used to develop predictive models. Surely, there are several possibilities we will be able to open”.
Conclusion... or maybe not
Neodata’s next step is to go on testing proposals and combine them to find the best possible solution for the Cookieless new world. At this very moment, Salvo’s team is testing the creation of Demand-side interest groups through its Audience Platform exaudi; also, they are testing on-device Auctions on their ad server ad.agio.
And they are making a proposal too: to develop interoperability flows among the different systems, to simplify and automate the process (for example, from exaudi to GAM, from Google Ads to ad.agio).
The Cookieless scenario surely is an intricate riddle, but with a good team the solution will come.
