RTB House’s Łukasz Włodarczyk sets out the landscape of digital ad targeting in a cookieless world.

We’re now approaching the final hurdle on the road to a privacy-first, cookieless industry. The options available to advertisers, marketers and publishers have already been well stated.

Alternative identifiers, such as email, offer a subsidiary switch from third-party cookies to new universal identifiers, group-based marketing can anonymise individuals while maintaining personalisation, and contextual targeting, driven by Deep Learning, can serve ads in the right context, at the right time, and in line with audience interests. And all without any third-party data at all.

But each of these approaches come packed with their own strengths and weaknesses, and each requires varying time frames to be applied successfully. The long and short of it? Advertisers need to combine long-term and short-term solutions to be successful and show commitment to privacy-conscious consumers. In this article, I will explain which solutions are most effective and how they can be combined.

The short – Alternative forms of data

For its entirety, programmatic advertising has relied on third-party cookies to accurately target different audiences and individuals with advertising. This data was incredibly valuable to publishers and advertisers, providing universal cross-site identifiers that allow them to build behavioural profiles. However, as we approach the advent of Google’s removal of third-party cookies, email, as an alternative identifier, is becoming very popular among advertisers looking to maintain similar practices. This is understandable – it’s natural to want an easy transition.

But what constitutes alternative data?

User profiles, often required or promoted by publishers, are referred to as first-party data. Because it’s consented to, publishers can – in theory – utilise this data to help effectively target users with ads while working within their own publication.

Equally, and perhaps an even more valuable source of data, the loyalty card model used by retailers is quickly becoming a highly accurate alternative. There is a far more appealing value exchange between the user and the retailer in this situation – the consumer will gain access to better deals and discounts on their favourite items. In return, the retailer gets accurate and consented purchasing information assigned to an email address.

However, for many people, the use of this data remains a potential problem if it’s used entirely on its own. Users will rightly question whether their data is being sold off to unknown companies and will likely demand increased scrutiny and control. It has the potential to act as a pandora's box of privacy, resurfacing the efficacy of targeted advertising post-cookie. If this form of data is overused, it’s only a matter of time before further legislation comes into play.

The medium – Group-based targeting

Often referred to as a ‘cohort’, a targeted group is one that anonymously assigns users to groups based on their interests alone. Google Chrome’s recent Topics API announcement is set to work on this premise, assigning users with five topics within a set of 350. Each topic is then mapped to the website’s hostname. The idea behind this system, and group-based targeting more generally, is to make the interests of users the defining factor for targeting.

The key benefit of these interest groups is the ability for brands to anonymise their audience, significantly improving privacy while helping target relevant ads. Once again, to use Google’s recent announcement, FLEDGE API is designed to use first-party data for advertisers and publishers to create group targeting, which is more anonymised and privacy preserving.

While the ad tech vendors cannot recognise the user as an individual, the ads rendered on their browser will match their interests.

The long – Contextual targeting

Contextual targeting as a method of advertising is already 20 years old, however, it has been revived by Deep Learning technology, increasing its efficiency and scale. Contextual targeting doesn’t rely on third-party data to follow a user across the web and can, instead, understand the context best suited for the advertising campaign. It achieves this by analysing keywords and other specified information related to the publisher’s content. In fact, the introduction of Deep Learning has significantly expanded the capabilities of the solution.

The Deep Learning algorithm reads specific signals — such as a website’s URL, content category, text, images, and video — to understand the contextual relevance of each page. The level of autonomy and self-learning possible with Deep Learning means the algorithm will only continue to improve its targeting capabilities over time, making it a primary, long-term solution.

Finding balance with a combined approach

The weakness behind each of these solutions is overreliance. If advertisers and publishers put all of their money and focus into a one-size-fits-all approach, they will risk the very same problems that are now associated with third party cookies. The best strategy is found in balance – combine each approach to maximise the effectiveness long term.

The sources of first-party data mentioned in this article can still be used effectively without compromising user privacy. The consented data, for example, can be used to more accurately create targeted groups from which to advertise, linking a user’s favourite reading and viewing topics to larger interest cohorts.

If contextual targeting – driven by Deep Learning – is then applied to this combination, the accurate interest group information provided can be merged with expert keyword and image analysis on-site. The level of insight possible from this and the ability for the algorithm to increase its accuracy over time make this method equal to, if not better than, current practices using third party data alone. Ultimately, harnessing modern autonomous technology is the key to unifying these approaches. 

Wasting hours on the hunt for a single solution is an approach that belongs in 2020. It’s time to combine the most effective solutions into one ultimate strategy and pave the way for a new era in digital advertising. Instead of an industry bereft of accurate data in the post-cookie world, I argue a combination of long- and short-term strategies will lead to an industry with more innovation than ever.