Advertising Digital Publishing
8 mins read

Looking towards a cookieless future for publishers: Part 2

In this report, we take a closer look at the current advertising ecosystem, as well as examining how the industry is responding to the cookie challenge. We also ask industry leaders their opinion on the steps publishers can take, as well as which emerging solutions, or combination of solutions, might help publishers pave the way forward in this cookieless future.

Click here for Part 1 of this story.

With a risk of collaborative solutions being hampered by privacy legislation, is there a way of using the data without breaking compliance? Arguably, edge computing provides a perfect solution. 

Enter: edge computing (the era of zero-party data?)

A relatively new technology, edge computing collects and analyzes behavioral data on the user’s device to serve them targeted ads (rather than sending the data to be processed on a server as with most behavioral targeting). The advantage is that it minimizes the data leaving the user’s device, protecting their privacy and remaining compliant, while also preserving personalized ad space. Edge computing also circumvents restrictions on the amount of data allowed to be sent into the bid stream, either through legislation (e.g. GDPR) or browser technology (e.g. ITP).

Edge computing is unique as it doesn’t use IDs but rather an abstract description of the user identified across segments (e.g. ‘baking’ and ‘dessert recipes’). When an ad request or bid request is made, the abstract description is shared with the SSP or DSP respectively. Although ad tech is a nascent use case for edge computing (time will tell if it delivers on the promise of speed, accuracy, and impact on CPMs), it does seem to offer the optimum balance of user privacy and publisher insight.

Claiming a 100% match rate, Permutive has established itself as a leader in edge solutions for many publishers including Dennis, Business Insider, Conde Nast, Immediate Media, and The Economist. As David Reischer, Head of Product at Permutive, explains: “user data, by design, is protected because it never leaves the device. This means a publisher can process their first-party data in a world without cookies and with increased government regulation – even under its strictest interpretations. Permutive uses data about authenticated users (10% of visitors) to extrapolate and produce a predictive model about the unauthenticated visitors (90%).” Using Permutive, publishers can collect for instance 20 or 30 customized real-time data points, or ‘events’, about visitor behavior in a single anonymous pageview (e.g. article title, author or category).

Meanwhile, ID Ward is a privacy-by-design initiative partially funded by the EU, not only due to its alignment with the EU’s privacy and sovereignty agendas, but also its vision of protecting the diversity and financial sustainability among (smaller) European publishers. Pushing the boundaries of innovation in edge computing, ID Ward’s solution goes one step further by not limiting publishers to domain data. The platform uses ML to offer publishers insights about their audience cohorts (user churn rate, apps used etc.) and complements Google Analytics, measuring traffic to and from a publisher’s website. But, at the same time, it uses edge computing to keep the data anonymous.

As Fosci explains: “We manage the whole data lifecycle from collection to the creation of the anonymous audience segments. And in all that, we are not preventing publishers from also viewing and exploring their own data with their own tools. But we can give publishers the benefit of targeted advertising with full compliance; we also can give them anonymized information about their audiences, which they can use for subscriptions. The engine that’s powering personalized, targeted advertising is the data that sits on the device, so it’s zero-party data in that respect”.

A potential drawback is that to achieve a swing of market power, publishers would need to ‘fill in the gap’ by modelling and segmenting users. This would require substantial investment, perhaps negating any server cost savings. In addition, to be scalable across different publishers requires agreement on a common taxonomy for the abstract description of the user. Also, the valuable nanoseconds saved by removing the need for ID matching might be negated by the solution demanding more of a user’s hardware resource – on a device currently less powerful at data processing than a DMP or CDP – potentially resulting in latency. In contrast, as Pranay Prabhat, Senior Director of Digital Ad Technology at NYT suggests, “server-side processing is especially important to keep our front-end web and mobile apps lighter and faster”.

However, as general device technology evolves, edge computing will definitely become a more scalable solution. Interestingly, alongside Permutive’s and ID Ward’s DMP solutions, a new privacy-first browser has also emerged, from Brave. Its Brave Ads program uses granular profiles controlled and maintained by the user and incentivizes users by offering 70% of the revenue shares in BAT (Basic Attention Token), a reward point redeemable within its loyalty platform.

Beware of behavioral, count on contextual

With around 86% of programmatic ads in Europe using behavioral data, conventional wisdom is that behavioral targeting improves campaign effectiveness, and therefore, advertisers pay more for behavioral data. However, there is little empirical data showing that behavioral targeting alone significantly improves publishers’ revenues. A recent study assessing the impact of cross-tracking ads on publishers found only a 4% gain in revenue ($0.00008 per ad). With the financial and reputational cost of compliance potentially outweighing any small gain in revenue, contextual targeting may be a better bet.

Before the advent of the web, behavioral advertising didn’t exist; all targeted advertising (TV, print) was contextual. Combined with programmatic trading, contextual ushered in the dream of one-to-one advertising and is still a viable method of monetization today. Contextual relevance can make ads more engaging and memorable, and can also prove more effective than behavioral for a geographically local campaign in terms of Cost per Acquisition (CPA). As per the earlier car vs fridge scenario, contextual ads can be more engaging, as they more closely match the user’s real-time intent, rather than perceived intent based on how they behaved two weeks ago.

Looking beyond a sole reliance on first-party data, Reischer says: “By embracing the anonymous web, publishers can also reach the audiences they want in a contextual, relevant, and timely way. It will mean moving from a world of one-to-one marketing to one where audiences, not individuals, are key”. Meanwhile, Cackel explains: “Contextual targeting doesn’t require the use of cookies and is definitely not a plan B. Teads has been using it across industry verticals and markets, and successfully provides similar media effectiveness to clients. Contextual targeting works harder with quality content, and this should be a key focus for publishers to thrive and differentiate themselves”.

Modern-day contextual targeting – which has evolved into ‘contextual intelligence’ – uses AI and ML to achieve:

  • more accurate and granular categorization of content
  • more dynamic content at the page and post level (e.g. ‘’ could be ‘news’ at the top level, but ‘’ could be ‘news’ and ‘technology’)
  • capabilities across more languages
  • sentiment analysis and mitigation of mismatched ad placements to improve brand safety for advertisers. For example, previously a recipe website might have excluded the words “hash” and “herb” to avoid association with content related to illegal drugs, thereby making it difficult to reference “hash browns” and “herb-cheese omelets”. Sophisticated contextual solutions also go beyond text to recognize images, video, and audio.

British startup Illuma Technology has developed proprietary AI to improve contextual targeting by reading live contextual signals that are driving interest and awareness of brands on the web. The platform then uses these to expand its search and identify new prospects for relevant ads in real-time, which it then serves programmatically. As a recent Kantar study reveals, Illuma’s technology grew ad awareness by almost 12% (65% above Kantar’s UK norm), and brand affinity (“an important yet difficult metric to shift”) by almost 15 percentage points.

Following a period of heavy focus on behavioral targeting, it seems an AI/ML-supercharged contextual comeback is presenting a major advertising opportunity to fill the void. Other contextual intelligence solutions and initiatives include Beemray, The Ozone Project, and Comscore.

Bridging the gap

In what’s turning out to be an exciting era for the online ad industry, we’re seeing the emergence of a new technology known as ‘moment targeting’. As Peter Mason, Co-founder of Illuma, explains, moment targeting fuses behavioral targeting with contextual, with AI forming a bridge between the two. It uses real-time audience insights to prospect contextually, reaching people when they are most receptive. “Advertisers benefit from the scale and anonymity of contextual while maintaining the relevance of behavioral targeting. Moment targeting delivers results across a range of KPIs, from brand-based metrics such as reach, video views and brand uplift, right through to performance targets such as form fills and sales”. Importantly, this type of targeting is unaffected by browser changes and doesn’t use personal data. Instead, campaign moments are defined using live page data, making moment targeting a compliant and future-proof option for marketers in the privacy-first world. 

Supplementing with subscriptions

The consensus among industry leaders is that regardless of other solutions being adopted by publishers, renewing focus on building a first-party subscription base is crucial. As Mason advises: “publishers – particularly small to mid-tier – should think about growing their first-party audiences and learning as much about them as possible. Simple steps might include gating online content with passwords and driving newsletter subscriptions”.

Success stories include The Spectator, which more than doubled conversions in three months and is forecast to reach 100,000 subscribers in 2021. The publisher achieved this by streamlining the entire subscription process (from login to payment to checkout) with the help of Piano, an e-commerce optimization and analytics solution providing out-of-the-box templates and segments that allow the creation of customer journeys at speed.

The advice from industry leaders is to offer a value exchange and make users feel like an exclusive member of a club. To tailor content to subscribers (or prospective subscribers), it should be localized, personalized, publisher-specific in terms of perspective, relevant to daily life, delivered at a regular cadence, and optimized to provide the best user experience.

As Erismann stresses, “while it is part of building a longer-term strategy, publishers need to be looking in-depth now at how they develop authenticated audiences, for example by an individual registering on a website, signing up for a newsletter or accessing a community forum with their email address. At the core of this process, publishers need to look at the value exchange they are entering into with the customer, which means offering meaningful content, experiences and more”.

Whereas publishers used to believe that readers wouldn’t pay for content, most publishers (78%) charge for digital access, with the majority (80%) offering a limited number of articles, and over half (57%) limiting access to five articles or fewer per month. Therefore, publishers should consider using a metered model rather than a hard paywall. However, as Hogg warns, “walled gardens may not be viable for small and medium-sized publishers [that rely more heavily on pass-through traffic]. Publishers that do have strong subscriptions are leaning towards mixtures of identity and context”. White adds: “Not all publishers have the brand recognition necessary to command subscriptions, and a barrage of logins will diminish the browsing experience. This is why Quantcast focused on developing Permisio”. Other vendors offering subscriber analytics for publishers include Chartbeat, Oracle NetSuite, and SimilarWeb.

Vertical ad networks (niche networks for niche titles?)

In the late noughties, vertical ad networks became popular among publishers as a privacy-friendly way of dealing in ad inventory for specific topics to target an audience based on their interests (technology, automotive, travel), as opposed to a horizontal ad network dealing with a wide variety of inventory. Tapping into a vertical ad network is a privacy-friendly way of targeting, as you’re focussing on the content rather than the user. However, it’s only viable if your content falls into one of a few specific niches; otherwise, you may see CPAs increase as the ads become less relevant. Nowadays, most networks seem to have been replaced by other technologies, or been built on top of horizontal ad tech, although some still exist today, such as GourmetAds (food and wine) and its subsidiary HealthyAds (health, fitness, and medical).

This article is an extract from our free to download report, How Publishers Can Swap Out The Cookie Jar.

Hazel Broadley,
Author, How Publishers Can Swap Out The Cookie Jar