Google Hands Privacy Advocates Another Dartboard with “Topics”
Which side are you on?
Are you a surveillance advertiser that tracks helpless consumers while you force them to buy things they don’t want?
Or are you a privacy hawk who thinks it’s best if Floridians get ads for snow pants?
It’s not quite that simple, but it’s starting to feel that way as the dust settles from Google’s latest attempt at quelling our post-cookie advertising fears. “Topics” — the next evolution of the failed FLoC approach for balancing anonymity and personalization in advertising — seems unlikely to sway the opinions of those who’ve chosen sides on the privacy debate, and that’s not because it’s a bad idea. Let’s unpack what’s happening and where things might go from here.
What are the key details for Topics?
Topics is similar to FLoC in that its core mechanics entail internet users being assigned into cohorts to guide the trading of ads. These cohorts are meant to be descriptive enough to entice advertisers to use them for targeting without enabling fingerprinting, x-device tracking, rich user profiling, resolution to offline IDs, or other privacy no-nos that sent us down this road of cookie deprecation in the first place. Advertisers access the topics through Google’s Topics API, quickly reviewing a given Chrome user’s list of topics prior to deciding whether to bid on them or not.
Generally speaking, the effectiveness of any cohort targeting approach is dependent on two factors: 1) the granularity of the cohorts and 2) the composition of the audience within. With granular cohorts, precise groups of people can be specified for messaging. And with a clearly defined audience composition, advertisers can deploy these cohorts with confidence that they’re getting a meaningful audience target for use in advertising.
In terms of granularity of cohorts, we’ve gone from tens of thousands of potential FLoCs to just a few thousand potential Topics, starting with a modest 350 today. Similarly to FLoCs, these attributes represent a lightweight trail of web browsing history. If you want to target recent movers for example, you can raise your bids on auctions targeting Chrome users who’ve been to home improvement sites.
If 350 topics feels a bit blunt to you, wait until you actually see what’s on the list! It’s hard to imagine bland indicators like “Internet & Telecom/email” or “Online Communities/Social Networks” could have much practical use in advertising. Traditional ad targets like life-stage targeting and demographics are notably missing here. In fact the list looks like it was built more to satiate privacy advocates than advertisers, and we should expect this list to balloon over time if this product is actually going to drive ad revenue for Google.
One of the main differences between the two is how the cohorts are assembled — a chief concern noted by critics of FLoC. With Topics, the approach has been largely simplified: Whereas FLoC group assignments were based on observed browsing behavior and thus subject to potential machine learning bias that could lead to discriminatory advertising practices down the road, Topics is human-curated, and the topics themselves can be accessed and deleted by a given Chrome user. On top of this, 5% of added topics are deliberate noise meant to confuse ad tech companies using Fingerprinting tech to attempt to stitch together users across devices. So while Topics might not be targeted enough for many modern advertisers, it certainly reduces the ability to track users and gives users more options to opt out relative to FLoC. As such Topics should feel like a win for consumers and privacy advocates alike.
What’s the industry reaction?
In reality, few privacy advocates seem convinced this is a meaningful step. Nonprofits such as EFF and Public Knowledge are on record calling out Topics as susceptible to the same cross-device tracking that FLoC could have enabled. And in terms of consumers, it’s still unclear how much they fully understand the details of what’s going on in the tracking debate, with reports suggesting that half of US consumers happily accept all browsing cookies despite over 80% expressing concern over how web sites use their personal data. (Worth noting: regulators are very much active in trying to help us resolve this dissonance.)
Advertisers, for their part, seem mostly unimpressed, with some clamoring for “more innovation” and saying Topics was “not designed to be effective”. When compared to existing ad tech solutions for targeting that are built on rich online and offline behavior — or even some of Google’s own other proposals — Topics certainly feels a bit milquetoast.
As such, the designation of equal parts “too revealing” for privacy hawks yet “not addressable enough” for advertisers seems to be unshakable for any of Google’s proposals, to date or in the future. We’ve already picked sides in this one — those who want more effective targeting vs. those who want better privacy protection — and nudging the solution in either direction is unlikely to drive us to an acceptable resolution for all parties.
As easy as it is to forget: most digital ad impressions aren’t on Chrome anyways, and your brand’s Chrome’s footprint is likely to slowly decrease as audiences shift from web browsers into connected TV and mobile. While Google’s proposals deserve evaluation and scrutiny, it’s unlikely you’ll need to bet the farm on them anyways. Rather, Topics — or FLoC, or whatever we end up with from Google— will be one of many solutions next to 1st party data, dynamic geo-targeting, 2nd party publisher data, contextual targeting, and other durable targeting solutions that brands can deploy now and into the cookieless future.
Don’t feel like you need to obsess over whether Google nailed it or not this time — instead, take comfort in all the other things you can do to drive customer growth without even having to say the G word.
Sticking to vegetables,