Digital Publishing Guest Columns
5 mins read

Why AI is the secret sauce for publishers’ cookieless environments

As we steer toward a rapidly approaching post-cookie world, one thing we hear again and again is: this will be great for publishers. Savvy publishers will see this as an opportunity to monetize their first-party and contextual data. Inevitably, there will be both advantages and challenges to this approach.

Publishers are going to have to compete with the ever-growing platforms for ad dollars, and they will be coming at this from a serious disadvantage from a logged-in user perspective. Aside from a few giant subscription-based brands, trying to take on a global social platform in a battle for who can collect the most email addresses is not even close to a fair fight for the average publisher.

What do I mean? Well, the conventional wisdom in the market is that as long as publishers are able to collect loads of data on their users – they won’t just be fine – they’ll find themselves in a new advantaged position. This seems simple enough: implement a registration requirement, and then just gather as many email addresses from users as you can. We’ve already seen this strategy work for The Times, The Telegraph, The New York Times, The Wall Street Journal, the Financial Times, and….well, that’s about it.

In reality, when we talk to digital publishers, even large, well-known brands, at best they expect to be able to convert roughly single-digit audiences to registered, logged-in users. Outside of pure subscription products that provide access to content that helps readers grow and/or make money (such as within their profession, or as investors), or delivers them pricey, exclusive entertainment – most consumers don’t want to create dozens of accounts to surf the web. Thus, it seems clear that going forward, we as an industry need to find a solution that works for the other 94% or so of web users.

As the Cookiepocalypse looms closer, we’ve seen varying degrees of enthusiasm for the establishment of an alternative universal identifier, one that pools information from hundreds of thousands of publishers. Others are pitching a set of identifiers that are interoperable, while contextual ad products have their devotees as well. And of course, we can’t forget artificial intelligence (AI). As the often-repeated theorem goes…surely AI can turn publisher data into gold.

The truth is, it’s unlikely that the digital ad world will coalesce on one solution – a product or data point that single-handedly achieves what cookies were able to, while also replicating mobile IDs and keeping regulators at bay.

There are many pros and cons to all of these potential solutions – shared email addresses feel like a ticking privacy time bomb, while contextual products can come across as imprecise and are not simple to activate at scale. Ultimately, it’s more than likely that we’re entering a multi-signal world – one with implicit trade-offs and fewer sure things – but potential better outcomes for all.

Certainly, brands and publishers should strive to collect as much user-consented first-party data as they can. There’s little doubt that this will continue to be an extremely valuable commodity in digital marketing for the foreseeable future. Yet for the vast majority of the trusted open web, publishers and advertisers will need to employ dozens if not hundreds of data points to effectively connect with customers. And yes, AI will be at the center of this. Perhaps not in the way you might think.

It’s true that publishers have more information about their audiences than they are given credit for, but even for large web brands, they can rarely tell much of value about a user based on a few site visits – considering that the average session consists of only 1.5 page views. That’s why, to fully unlock the potential of contextual data, behavioral history and interest-based reach, you need a cross-web solution. Such a solution must be based on not dozens of visits, but trillions.

Case in point. Think of a person who visits a food site. Without ‘knowing’ who this person is, that food site might serve an ad related to cookware, or a particular type of cuisine. However, that particular user might be searching for healthy meals because they are trying to lose weight, or because they have particularly unique dietary needs.

No one web publisher could be expected to have access to or understand this information – even if that site had plugged in the latest cutting-edge analytics product or single-use AI offering.

What’s needed in a situation like this is a map of the internet, which is a massive data and tech challenge that warrants a comprehensive, AI-based solution. This is something realistically only a few companies can even claim to have a shot at building.

The good news is, the technology is there. While the advertising industry has had products that scan web pages for keywords for decades, AI is rapidly surpassing this blunt technique. Increasingly, AI can understand content and sentiment through advanced natural language processing and image scanning. This can be done at scale at a level that was unthinkable even just a few years ago. However, these tools must be fed. The more AI collects such data, the smarter it gets. Over time, AI and machine learning can help put content pages into classifications that are far more valuable than ‘golf fans.’ So it behooves the publishing world to get moving now.

In addition, beyond content and user ‘classification,’ the modern AI industry is making huge advances in collecting and processing probabilistic data. If you’ve spent any time in ad tech, you may have a preconceived notion about what ‘probabilistic’ means. I urge you to forget it. Today’s AI tech isn’t about making assumptions, or chasing vague lookalikes, but rather making extremely accurate guesses which can provide marketers with a high level of confidence that they are able to connect with the right audiences. This isn’t the same as matching up deterministic ID sets, but it’s the next best thing.

And just like with content, probabilistic data can be accumulated and analyzed at scale.

If we as an industry pursue a full embrace of AI’s potential, the result could be a market where brands will have access to large volumes of real-time, consented data from a wide variety of sources. We’ll know just what that food site visitor is looking for. To get there, publishers are going to play a huge role in creating this reality. Which is why they need to be smart about which partners they choose to work with, and ask the right questions about these company’s investment level, technology prowess and data robustness.

Because not just any ad tech startup can pull this off. In fact, I’d expect a very welcome shakeup in that regard. Gone are the days when anyone with software could plug into a few exchanges and build a “me too” company. There simply aren’t a million vendors who can operate at this level of AI power. For marketers, this will mean less needless complexity and a reduction in the dreaded ad tech tax. As long as they choose wisely.

For publishers, AI can be truly revolutionary. It won’t just shift some leverage in their favor, it will help them monetize the vast majority of their users in ways that only the biggest Silicon Valley players could even attempt. That’s a much fairer fight.

Allan Tinkler
Head of Platform Business Development, Quantcast

Quantcast is an audience intelligence and measurement company headquartered in San Francisco. Combining machine learning, a privacy-by-design approach, and live data drawn from more than 100 million online destinations, Quantcast provides software, information and advertising services for marketers, publishers and advertising agencies worldwide. Founded in 2006, Quantcast has employees in 20 offices across 10 countries.