Can artificial intelligence solve data’s consent problem?
For The Drum’s Data & Advertising Deep Dive, Alexei Lee of agency Reading Room asks: will increasingly machine learning-reliant data products save the day, or is consent an intractable problem?
Can artificial intelligence help with the growing issue of user consent in data? / Pawel Czerwinski via Unsplash
There are darker days ahead for marketers. Imagine a dystopian land where website users lurk unseen in the shadows, and advertisers rely on nothing more than their wiles to decide who sees their digital ads. This is the cookieless future (or so we’re told), and it’s just around the corner.
Uttering the ‘C’ word can trigger this instant anxiety. Until recently, it was hard to imagine how the industry could survive such an upheaval of the status quo. Cookies allow us to target our ads, understand what people do on our websites, and measure how successful we’ve been. What’s the harm in that?
Thanks to murky attempts in the recent past to map and manipulate people using their online data, many people feel strongly about who has access to it. The arrival of GDPR brought welcome clarity on how data consent should be sought and significantly raised public awareness of data protection issues. The upshot is that we’ve become well-trained to un-tick in the appropriate places when that annoying box pops up on the screen (no matter how gray and obscure the button is).
With a sea change in public attitude toward data sovereignty, and the threat of users jumping ship, platform owners have been quick to respond. Apple was first to introduce changes this year, and others have followed suit, with Google committed to phasing out cookies by 2024.
This causes a headache for digital advertisers and, well, anyone who owns a website. Unhappy advertisers are bad news for big tech’s bottom line.
Login-based platforms like Facebook and Amazon benefit from users’ prior consent (and we mostly pay zero attention to when asked to read the T&Cs). But they still haven’t avoided messy cookie warnings. Meta forecasted a $10bn loss in ad revenue following Apple’s introduction of increased data privacy controls this year.
In the face of an increasingly consent-averse public, it’s a race to find ways to keep the advertising pennies flowing and the data business booming.
The new kid on the block: AI
Artificial intelligence (AI) has long been used by auction-based ad platforms such as Google to model how users behave, so that ads can be targeted more effectively. But the data privacy revolution has changed the game.
Emerging machine learning-based solutions such as Google’s Topics API and Google Analytics 4 promise to do more with less data, tracking and targeting individuals based on prediction, with reduced reliance on unique IDs to monitor users’ interactions.
AI not only helps to ‘fill the gaps’ where users have not allowed data to be gathered, but makes marketers able to track users’ full journeys, regardless of device (which Google’s current analytics offering can’t do).
On the face of it, Topics API and GA4 allow people to keep control of their data, but advertisers and website owners still get to do their thing. And with GA4 you get the bonus super-power to predict what they might do next (be that purchase propensity, basket value, or drop-off rate).
Everyone’s a winner, right? Possibly. Possibly not.
Dig a little deeper into this tech, and the apple cart starts to wobble.
What happens if no one consents, ever?
We now have a deep distrust in organizations and their ability to manage our data; this mindset will be hard to change.
Regardless of how many times Google and Meta tell us how anonymously our data will be treated, if non-consent is the default, we’ll keep it that way.
There’s unlikely to be a big regulatory departure from the basic principles of GDPR any time soon. How accurate will data modeling be if there’s hardly any new data to model from?
Is this just stalking under a different name?
Google’s Topics API gleans interests from a user’s browser history, and only shares broad topic data with advertisers to guide targeting. This seems less invasive, because no personal data is passed to a third party.
But gaining data consent is just as much about perception as technical reality. Will the layperson see monitoring of their browsing history as any less ‘big brother’? If not, users are still likely to seek out the ‘off’ button in their browser settings.
Is it all about context?
Another emerging AI-based approach is contextual ad targeting: AI is used to analyze the content on a page to decide which ads are most relevant to serve to the user. Sounds newfangled, right?
It’s not exactly innovation at its cleverest; more like the digital equivalent of placing a billboard ad about cars on the motorway, but done quicker.
That’s not to say context doesn’t get results. TV and Radio ads have been cookieless for decades but still deliver effectiveness. However, with one ad platform recently hailing a 3% increase in purchase intent as a big win for contextual targeting, you may be better off with the motorway billboard for now.
With the technology still nascent, the jury is out on whether AI will be the savior marketers are seeking to cope with a cookieless future.
There are some green shoots with the arrival of Google’s GA4, but consent is still front and center of the data sovereignty revolution and that’s unlikely to change. With trust in big tech at an all-time low, the future of digital marketing will be shaped by people, not platforms.
For more on how the world of data-driven advertising and marketing is evolving, check out our latest Deep Dive.
Content by The Drum Network member:
Welcome to the home of the down-right inquisitive.Find out more