Google has finally come out and said that FLoC (Federated Learning of Cohorts) is dead, or perhaps evolving, into Google Topics – an anonymous, browser-based, tally-based system to identify what you may or may not be doing across the web on a weekly basis. On the surface, it sounds like an interesting proposition for how to deal with users of Chrome’s data, while also providing them with a semblance of privacy as to who the user actually is, along with the acknowledgment that they don’t want to track them forever by wiping the data on a weekly basis.
What do I believe the goal of Topics was for Advertisers from Google?
The goal of Topics is to provide brands and advertisers with a coarse-grained selection of interest segments over an epoch of what a page visitor may be interested in. It is Google’s compromise of some layer of added anonymity for the consumer, while also trying to provide marketers with some type of insight into the context behind the user’s visit to the site.
So, how will Google Topics work?
Google Topics will be an API to which advertisers can connect, and have users’ Topics – or interest segments as I’ll refer to it from here on – returned back to them. This will provide advertisers some insight into the contextual behavior of the user, and what types of sites that they have been visiting over the past several weeks.
For the chrome users only, Topics will be embedded into a browser update coming in the future, from which Chrome will begin tabulating all of the sites that the user visits over the course of a week. These sites will be segmented into an interest taxonomy of approximately 350 at launch, that may grow to over 1,500 over time. Throughout the week, these tallies will increase, and the user will receive a ‘top 5’ interest segments, along with a randomly selected interest segment from the list of 350. I know, this sounds a bit strange, as if you are a travel enthusiast who is spending the week looking for your next vacation, you could also get Hedge Funds as one of your top interest segments, throwing off how a website or brand may view you, and the interests that you have.
This data is then sent back to advertisers/websites from which they can use it to customize the experience to the user’s interest segments on their site (if it is relevant) or through follow-up communications. This is somewhat complicated by the fact that the advertiser can only receive this data about the user once every three weeks, so if their browsing behavior changes the next week after say they’ve booked that vacation, and now they are looking for insurance, or maybe a new car because their lease is coming up, they may still be followed with those travel advertisements for a longer period of time. Additionally, an advertiser won’t receive all 6 of these interest segments, in fact, they will receive anywhere from 0 – 3 (which could contain that random one don’t forget), so as a consumer, we can all expect to have much less relevant experiences coming from brands.
What does this really mean for marketers’ interest in performance?
The ‘may’ be interested in is the discussion point in my mind for Topics, as the design of the product is that while the user’s browser will be tracking them throughout the week (there is a reset after a 7 day period, or they can clear their cache at any time to reset it), there are several concerns I would have about the application of it:
- Epoch – What period of time is this truly scaled over? Given that a brand will only be able to receive the interest segments once every three weeks, there is the chance that the visitor has already fulfilled their desire that drove a high tally of the interest segment that the marketer receive. This will lead to the wrong message being displayed to potential customers.
- Not Receiving an Interest Segment – Google’s documentation states that marketers will receive 0 – 3 interest segments, but what happens if you’ve designed experiences based on this data, and you do receive 0?
- Randomization – There is a 5% chance that a random interest segment will be returned to the brand. A 5% chance in randomization can also be viewed as a 5% waste in personalized experiences, contextual advertising, or the marketer’s offering. It also makes it much harder to have confidence in what marketers are receiving, and how their confidence intervals are statistically measured.
Overall, marketers will need to use this data as a contextual overlay, but not as a main driver or data point of who that user is, or what they are truly interested in. Layering this data with data points of what they already know about someone, and what they are currently doing on your owner properties as a confirmation point will be key to driving relevant experiences.
Should Topics be a primary part of your marketing plan in the future?
There may be some industry segments that Topics will be good for, especially for those who have minimal site experiences or first party data, but by and large I don’t think marketers will prioritize this product due to performance implications as I noted recently in an Ad-Exchanger Article – Ready Or Not, Topics Is Replacing FLoC – But Will It Be Useful For Advertisers? By and large, new products like Topics will continue to push marketers to form a relationship with their customer, acquire 1st party data, and make decisions off of it.
1st party data will be a marketers best friend and the primary route to being able to drive performance in the post cookie-world, where products like ‘Topics’ will leave performance marketers stretched to provide week over week, month over month performance improvements. Marketers need to begin working more closely with Publishers (Google, Facebook, Hearts, Comcast, etc.) and a match between their customer base vs. relying on low cost remnant inventory, or ad-tech that was previously able to track users across the internet. The Marketer, with their customers’ data, has the opportunity to be a lot more picky with who, and how, they engage, pushing for transparency and performance over just connecting with who could be their customer. Customer and transaction data is what all publishers will need to match based on, and what ad-tech is going to be desperate to get their hands on to stay alive in the coming years.
Ready to learn more about how Tealium can help you take advantage of your 1st party data? Speak with us today to learn how we can help your specific business.
