Google ads VP on future of data-driven advertising: ‘Prediction will supersede precision’

The tech titan is at the forefront of the mission to innovate new advertising solutions that promise effective targeting and measurement while respecting consumer choice and privacy by inhibiting user-level tracking. Its Chrome and Android teams are working alongside its advertising division – and collaborating with external partners and regulators – to help deliver a future of advertising that puts consumers first.

Since the 2019 rollout of Google’s Privacy Sandbox – the far-reaching initiative that aims to develop these new methods and standards – approaches have evolved. The company abandoned its original Federated Learning of Cohorts (FLoC) framework, which grouped users into ‘cohorts’ based on their interests, in favor of a new model called Topics that assigns specific topics to users based on their browsing behaviors. Its privacy-safe retargeting framework First Locally Executed Decisions over Groups Experiment (Fledge) has been rebranded as the Protected Audience API – a tool that aims to help advertisers retarget using first-party data when third-party cookies are sunsetted. Meanwhile, Google is pouring investment into predictive approaches as the AI wars heat up.

The efforts have seen mixed responses among advertisers and privacy advocates.

Here, we quiz Dan Taylor, Google’s vice-president of global ads, on how the company is addressing the challenge of balancing consumer privacy with advertiser demands.

Do you think consumer awareness about data privacy is growing? What does this mean for the advertising ecosystem?

Yeah, people are certainly more aware and more concerned about what data is being collected, how it’s used and especially who it’s being shared with. And that’s really driving a lot of the trends we’re seeing in digital advertising today.

We have some research recently that shows 80% of people are concerned about the state of their online privacy and almost half are starting to turn away from services due to privacy concerns.

Now, at the same time, people do like content that’s customized to them. That same research said 74% of people only want to see ads that are relevant and useful to them. Obviously, services like YouTube, Netflix, TikTok and your favorite news publisher are often built on recommendation engines that deliver personalized content.

So thinking about both this concern that consumers have about their privacy online and their preference for customized content can [seem] like a contradiction. But I think it’s an opportunity. The rich and diverse content on the web, where users can access a wealth of information for free, is enabled by this ad-supported business model, where publishers and content creators get paid for their work by helping businesses reach really relevant audiences. So if consumers are losing trust in that model, that whole cycle starts to break down. So what you’re seeing in the industry is [a shift toward putting] user trust at the center of it all.

How should marketers – whether brand- or agency-side – be approaching that goal of achieving a balance between delivering relevant ads to audiences and preserving consumer privacy?

The state of online privacy requires new thinking as we shift from what worked in the past to what needs to work for a more sustainable future.

Brands that are not prioritizing privacy are at a disadvantage. Another recent survey [we conducted] showed that 49% of people would switch brand loyalties based on a privacy experience that didn’t go well – or one that did go well.

Advertisers who think the end of third-party cookies is some sort of future state are already behind. The landscape has already changed, whether it’s in Safari, other browsers, iOS with AppTrackingTransparency, the state of consent in Europe and more. So when I’m talking to those customers, what I say is that digital ad technologies are moving from this notion of ‘precision’ to ‘prediction’ to deliver marketing results at scale.

What do you mean by ‘precision’ versus ‘prediction’?

What we're seeing is an increased use of modeling and predictive-based targeting and analytics that help you deliver relevant ads to audiences when you don’t have individual user-level data.

So [advertisers will be] increasing the use of tools – like optimized targeting, for example, which takes a bunch of signals that aren’t about a specific user, whether it’s contextual [information] or a time of day and other signals that have historically performed quite well – to anticipate what a relevant ad would be for a given impression when you don't have that same level of signal.

On the other end of the spectrum, once a campaign has delivered, being able to model for conversions and for actions that you can’t directly observe – but have a reasonable understanding, through predictive analytics, that those actions took place – enables a marketer to deliver relevant ads and also get some actionable measurement without that level of precision that [third-party cookies provide].

Quite frankly, the third-party cookie space has gotten us to a place where consumers are concerned about that level of precision.

To clarify, is it your understanding that advertisers are losing precision when they shift to these new forms of targeting and measurement?

Precision absolutely can be lost.

So the mental model that needs to shift [among advertisers] is that you don’t need the level of precision that you see today that is raising all of these consumer privacy concerns in order to deliver a relevant advertising campaign.

So where you’re seeing shifts in terms of use of AI and use of modeling to deliver outcomes and measure effectively is the pivot that we see the industry making.

What are some of the principles and standards governing privacy-focused ad innovation at Google? And how has thinking in this area evolved?

There are a couple of guiding principles that we think about as we’re building these solutions.

First is building with a privacy-first mindset. This means [building products that are] secure by default, private by design and put people in control. Those are some principles that we’ve stated publicly. But in terms of examples of this that make that more tangible, we never sell consumers’ personal information. We don’t allow targeting on sensitive categories like race or religious affiliation or health conditions. We give consumers transparency about the ads that they see – whether it’s the ability to see right within the ad who the advertiser is, or within My Ad Center which gives the user the ability to customize the ads that they see. So our first principle is making the default experience one that consumers can have trust in.

In terms of the technologies themselves, another belief or principle that we hold dear is you don’t need to track consumers across the web and apps to get the performance benefits of digital advertising. That’s why tools like optimized targeting and conversion modeling are so important to advertising not just now, but in the future. And then [as far as] the Privacy Sandbox, the work being done there is guided by this principle as well.

One of the unfortunate side effects of blunt approaches like blocking third-party cookies without alternatives that work for the ads ecosystem is that it has led to a rise in fingerprinting and other covert forms of tracking that we see as worse outcomes for privacy. The reason we’re not interested in fingerprinting as a technology that we want to pursue for advertising is that it tracks people across the web, it doesn’t have transparency and it doesn’t have control. So that’s three strikes for us.

Then finally, a principle that’s really coming through in the Privacy Sandbox initiative itself is collaboration. Chrome's been clear from the start that the initiative [to shift from third-party cookies to new approaches] would be collaborative and iterative. We’re working with the World Wide Web Consortium and publishing details on its progress, whether that’s on its website or on GitHub, as well as collaborating with a competition and privacy regulator in the UK and expanding those commitments that we made [in the UK] globally.

Things like the movement from FLoC to Topics, is an example of where feedback from the industry led to the development of a new API. And similarly, you’ve seen names like Turtledove and Fledge evolve to Protected Audience API, again, reflecting feedback from the industry on the right mix of delivering results for advertisers and publishers, but also not compromising user privacy. I think the collaborative piece is really the most interesting part of the initiative.

What innovations or efforts in Google’s ads business are most exciting to you right now, especially within the context of privacy-preserving approaches to targeting and measurement?

Three big areas that we‘re invested in are first-party data, use of Google’s AI and new privacy-preserving technologies.

On the first-party data piece, customers [need to] really recognize the importance of investment in stronger relationships with their consumers – obviously, that has always been a good idea as a business owner, but in today’s digital marketing landscape it’s even more so.

[One part of that is analytics.] Google Analytics was built with privacy at its core, expressly for a world without third-party cookies. So features like consent mode, predictive audiences and data-driven attribution are built right in. I’d be remiss to not say that Universal Analytics is sunsetting on June 30, so we’re encouraging all of our customers to make the shift to Google Analytics 4 today.

With that solid measurement foundation in place, we’re seeing brands put first-party data to work to differentiate their marketing campaigns and differentiate their monetization from a publisher perspective. Tools like enhanced conversions and Customer Match help create more clarity in terms of measurement and audiences that you want to reach where you have a trusted consumer relationship.

[We also have] new innovations like PAIR, which stands for publisher advertiser identity reconciliation, which represents a secure way for advertisers and publishers to connect audiences that they have in common via data encryption. Another is publisher-provided IDs and signals, which help website and app owners showcase the value of their unique first-party audiences to buyers in a way that scales but doesn’t rely on cross-site tracking.

Using AI to adapt to today's changing privacy landscape has proven to be an effective tool as well. Performance Max, [a tool that enables performance marketers to access all of their Google Ads inventory from a single campaign] is really the best example of how we’re using Google's AI to power results for marketers and really move at the speed of the consumer, as opposed to pre-planned budget silos. But in addition to delivering 18% more conversions at the same average cost per action, it turns out that predictive bidding audiences and model measurements kind of smooth out the bumps, if you will, in a fragmented landscape and are a win for privacy… [in helping] to deliver relevant ads where you have that signal loss.

The last piece is privacy-preserving [tech]. We already talked in concept about the Privacy Sandbox APIs. But last month, on the ads side of the house, we published the results of a recent experiment we did in serving interest-based ads with a combination of privacy-preserving signals – while suppressing third-party cookies – to see how those audiences would perform. And we saw that those campaigns maintained a pretty high degree of fidelity relative to third-party cookie-based performance, whether that was on scale, advertising, ROI or even relevance to consumers. We found that pretty promising and we’re going to keep testing the API throughout the course of the next year.

How are broader industry changes like the rapid adoption of AI and web3 technologies impacting how Google is approaching ad targeting, measurement and data privacy?

Ensuring consumers can feel safe online and in control of the ads that they see is a big part of how we’re trying to address user concerns with their privacy online. So when I think about tools like AI, we’re at a healthy inflection point for the industry where user privacy becomes a key part of the marketing decision-making process and also a key part of the technologies that you end up using.

Being able to apply AI to targeting through tools like optimized targeting or bidding or even generative AI that helps you reach a consumer with a relevant ad regardless of the surface really helps build a better advertising experience. Tapping into these predictive technologies to deliver an ad without the need to track an individual user online is having a great privacy benefit.

It’s an interesting intersection where consumers are more concerned about sharing their data and platform and regulatory changes are fragmenting the legacy technologies, but AI is reaching a point in its evolution where marketing tools can do more with less data and deliver relevant ads and actual measurement without that need to track with a such a level of precision.

On the web3 side, I think that is an interesting space and it’s definitely something that our Cloud team in particular has been looking at in terms of how they can support our customer needs on open source platforms. So we’ll be watching that space closely.

Another factor here is regulation and enforcement. A growing number of US states are implementing GDPR-like comprehensive privacy laws; the Federal Trade Commission is cracking down on what it calls ’commercial surveillance and lax data security practices’; data privacy authorities in Europe are doling out big fines for privacy missteps; the list goes on. How does increasing regulatory and enforcement activity impact Google’s goal of balancing user privacy with business demands?

I was in Europe a little over a week ago and the level of privacy assurances that we can provide to our customers – in equipping them not only for regulatory compliance, but also for ensuring a good consumer experience – was central to literally every customer and partner meeting that I had.

So obviously, we want to continue to evolve our offerings in tune with what our customers need, but also, our priority is to ensure that people’s data is protected as they browse the web.

Thinking about the US in particular, we’ve long advocated for a national privacy law in the US. We see the state-level actions taking place, and we’re working collaboratively in every jurisdiction where those things are being developed. But we think that a national privacy law in the US is what most people want to see in terms of being able to protect consumers at scale and not have to think about different changes that might take place from state to state.

But of course, complying with evolving regulation is table stakes. The fundamental issue, in my view, is increasing user trust in the experiences online through sensible policies, alongside transparency and control for users.

We talked a little bit about some of the things that we do from a principles standpoint, but look, we never use the content that people create and store and apps like Drive or Gmail or Photos for ads purposes. And we don’t use sensitive personal information for ads purposes either. Those are sensible ads policies that a reasonable consumer should expect.

In terms of transparency and control, we’ve done a lot there. In 2009, we were probably the first company to allow a consumer to make changes to their advertising preferences, including whether to be advertised on a personalized level at all.

But we can always do more. So last year, we launched My Ad Center to make our existing controls more prominent to consumers. We also built on our advertiser verification program this year, by launching a new Ads Transparency Center where consumers can look up the advertiser that they’re interested in, or don’t want to see ads from any more, and see the creative that’s been served and make changes right there.

[We’re committed to] that combination of being a responsible steward of a consumer’s privacy when we’re building our product, but also being transparent about how data is being used and giving consumers the option to make choices and making those controls easy to find.

At a time when policymakers and legislators are thinking about how laws can be adapted to new technologies, our engagement with regulators like the UK’s Competitions and Markets Authority on the Privacy Sandbox is one example [of Google’s collaboration]. We’re very much leaning in to work with regulators about how they can build sensible regulation that moves at the speed of technology and isn't outdated at the moment that it's written.

How would you summarize Google’s standpoint?

The digital ads ecosystem needs to evolve – most everyone can agree on that – to be safer for people so that they feel protected. But it also still needs to be successful for publishers and creators so they can find quality journalism and the content that people love. That’s really important because an open web is critical to Google’s mission to provide access to information.

We think we can do this while ensuring people’s privacy is protected. And hopefully, the investments we’re making in AI, first-party data and privacy-preserving technologies are tools that the industry can embrace to make that shift.

Sign up for your
Daily Dose of
Marketing Insights

Sign up now

To read more from The Drum’s latest Deep Dive, where we’ll be demystifying data & privacy for marketers in 2023, head over to our special hub.