Privacy-First, Precise, and Scalable: Targetable Contextual Audiences Have Arrived, and They’re Here to Stay
Content often inspires us to do something, to think or feel something which resonates with how people perceive brands and marketing messages. Contextual audiences work the same way, looking at who shares a common interest to create targetable audiences of any composition at scale on the open web. There is no need for data beyond what is on the page, contained in the text, images or sentiment.
Imagine you’re sitting in front of your favourite painting at the National Gallery. People of all ages, nationalities, genders, and income levels filter in and out of view as they stop to appreciate the art.
In those moments, despite their differences, all the people who passed had a shared interest, a common passion which created an audience for the painting and emotional engagement for the consumer of the work of art.
Of course, few brands have as broad an appeal as Van Gogh’s Sunflowers, so contextual audiences need to be deployed wisely and with consideration for what the advertising is trying to achieve by association with the content.
Luckily, on the open web we don’t just have a single painting to capture interest from, we have millions of articles published every day, each generating varying levels of interest and connection.
By mapping these web pages onto a contextual graph that draws connections based on content and context, we can reveal clusters of interest that serve as targetable audiences and show how different pieces of content are interconnected.
Let’s say a luxury clothing brand wanted to target prospective customers. The typical approach would be to select a small site list of a few high end fashion titles. Then in the old model any data like personal identifiers, or soon to be obsolete third-party cookies would be applied to try to segment the audience into the luxury brands pool of potential customers.
Contextual audiences, on the other hand, work from interest outwards, allowing the creation of hyper niche segments that resonate with the luxury brands audience and with constant machine learning we can fine-tune these audiences to be even more relevant, precise and valuable without losing scale and ensuring maximum exposure to the luxury brands message and to potential customers.
This audience would not only be found on articles specifically about fashion but would pull in other content based on their affinity to the core interest.
The contextual graph might reveal a cluster of overlapping interest between wanting sustainability and ethically sourced clothes to broader topics like travel, dining out, or even seemingly unrelated topics such as finance or investing.
The luxury brand simply has to select their contextual audience and watch their campaign proliferate across whatever inventory is most likely to be seen by the target segment. And the process is entirely user-agnostic from end to end.
The brand’s prospect could pick up a brand-new device with no user data or logins saved on the browser and immediately be served relevant ads through contextual audience targeting.
Now, anyone who’s been around for some time in digital advertising knows contextual advertising is nothing new. So why has it taken so long for targetable contextual audiences to become available? The answer is simple: we didn’t have AI.
Contextual audiences demonstrate how AI can be used for good
AI hype is at its peak, capturing mainstream news attention like few other technology innovations ever have, for better and for worse.
We’re yet to understand the full impact it will have on digital advertising, and inevitably there will be some missteps along the way but in general we will see massive benefits for more effective advertising, less likely to annoy consumers and more likely to yield results.
Contextual advertising has utilised AI to automate the categorisation of text and visual elements for some time, and the same technology is being applied to brand safety solutions at source without the need to restrict access to quality content.
Today, interests evolve rapidly and our culture of immediacy has seen consideration windows shrink. Contextual audiences must be constantly updated to remain relevant and precise, with audiences constantly expanding, contracting, and overlapping.
To give an idea of the scale involved in this upkeep, Seedtag’s contextual AI, Liz, processes up to 80 million articles a day across our network of 8000 publishers.
But we can’t discuss the power of AI without mentioning its controversies, particularly around intellectual property and data scraping.
While Liz is only fed data from our network of publisher partners, other AIs have not been so selective, hoovering up and monetising data from sources that have had no say in the matter.
This is not just an issue with generative AIs such as ChatGPT, but also contextual solutions, which prompted the AOP to publish an open letter condemning the practice of unauthorised data scraping.
It's important that contextual audiences and the AIs that power them are deployed responsibly for the benefit of all parties involved. There’s a lot on the line here, because if we get it right, we can usher in a new era of future-proof, privacy-first advertising.
Connecting with your audience shouldn’t be a game of ‘Guess Who?’
Third-party cookies are on the way out, privacy regulations have restricted data portability, and Apple keeps cutting the connective tissue of targeting tech. The digital advertising industry is now at a crossroads: try to recreate the behavioural approach that drew backlash from consumers and regulators, or establish a new, future-proof paradigm that is above reproach.
Contextual audiences enable precision targeting at scale without cookies and without touching any user data. The only data gathered and processed are article context and non-identifiable website usage data from across the partner network, which even the most hard-line privacy advocate would consider appropriate.
And — as long as they are informed and on board — publishers gain the ability to monetise inventory for all users, known and unknown.
This approach doesn’t just tick privacy boxes, it’s more conducive to the diversity of our increasingly connected world. Targeting based on assumptions about who is interested in what, rather than letting the interest speak for itself, results in potential customers being ignored in favour of low-hanging fruit, simply because they don’t fit the expected demographic criteria.
The cookie-based era of targeting was like a game of ‘Guess Who?’, with brands trying to figure out who their audience is based on a patchwork of behavioural data and broad-brush traits.
Contextual audiences take the guesswork out of identifying potential customers, allowing brands to connect with prospects based on their active interests, at the exact moment they are most attentive. Instead of pigeonholing people into restrictive categories, contextual audiences let customers show you who they are, on their terms.