As advertisers eye end of cookie, Havas Media unveils AI-enabled planning methodology

“Without cookies, most of the ways we have traditionally targeted audiences and measured the impact of targeting an audience will become redundant,” says Laura Kell, chief data and product officer at Havas Media Network.

To meet that challenge, the media agency’s UK business has retooled its media buying platform, Converged, to operate without the use of cookies or personal data – instead using demographic or behavioral information gleaned from audiences to lead media planning.

• Marketers & publishers brace for impact as 1% of third-party Chrome cookies vanish

Media planners already had access to rich data on target audiences, Kell says. But the data used to decide which consumers to target with a given campaign was “rarely” passed along to staff executing a media investment. “Our planners might spend a month building a rich portrait of an audience, and then someone else buys ‘men in market for a car’,” she says.

Converged, Kell explains, links the audience attributes pinpointed by planners directly into digital media buying platforms. “Instead of a planner recreating the audience in Facebook or Google, we’ve built a model that maps those audience insights directly on to the platform,” she says.

The methodology allows Havas planners to create target audience cohorts across different platforms. Paul Bland, head of biddable media, says the agency’s tests have seen improvements in select key performance indicators – mostly view through rate (VTR), cost per mille (CPM), cost per lead (CPL), cost per click (CPC) and cost per view (CPV), though Bland says the tests represent the “full-funnel” – in 93% of cases.

So far, Converged is being used for media buys placed on Meta, TikTok and Google’s DV360. Kell says the team plan on adding further platforms later this year. “Our ambition is to take it to more traditional channels,” she says.

Kell says the underlying dataset is “one of the most comprehensive” created in the marketplace. Ahead of third-party cookie deprecation (due in the third quarter of this year), it draws on demographic, consumer, attitudinal and geographic data. (The machine learning algorithms used to sort that data are the AI part of its ’AI-driven’ proposition). Overall, the dataset contains 35,000 data points on 500,000 British consumers.

Given the depth of the dataset, Kell says, Havas is confident that its probabilistic approach (that is, making qualified guesses about consumer segments based on indirect data) will be more effective than a deterministic one that relies on digital identifiers such as cookies.

Eventually, the company hopes to expand that set to include several million consumers. Client-owned data, Kell says, could be a means of developing it further – and of adding deterministic insights into a predominately probabilistic approach.

“Cookies and first-party data give a view on people, but not the true understanding of why someone makes the decisions that they make,” Kell says. “People are not explained by looking at what they do online. They’re not explained by what they purchase. We wanted to build an understanding of who people are, how they think and feel, what motivates the decisions that they make, and then what do they do and what they buy.”

GroupM began a testing program with Google’s Privacy Sandbox back in November, while UK programmatic firm MiQ foreswore the cookie back in December. Havas’ platform is the latest agency answer to attempt to steer advertisers towards a cookie-free future. Though versions of Converged have been deployed across international markets since the turn of the year, Havas is pioneering the latest editions in the UK.

10 clients have already tested over 30 campaigns through the platform, representing a “significant investment” according to Havas – though a spokesperson declined to name any others apart from homewares retailer Homebase, or the cash amount of media spend those campaigns represented.

Currently, those tests are being conducted through Converged and Havas’ normal buying and planning process. Eventually, most buying will be done through the platform. At Homebase, the media agency has been measuring cost-per-lead (CPL) as its primary metric.

Lisa Tickle, director of digital and marketing at Homebase, said in an email that cookie deprecation had rendered its previous media strategy inert. “The move to first-party data… has made it increasingly difficult to match planning with the audiences we want to reach. Converged eliminates this challenge and has already yielded some very impressive results for us across a number of our product categories and channels.”

“We were really pleased with the results from that content. Cool. It was because it was a business outcome. And we’re able to deliver a big uplift in that outcome,” adds Kell.

As well as increasing effectiveness, the company hopes that the platform’s streamlined process will save Havas media staffers time. Though its rollout has been prompted by the slow exit of the cookie, Bland hopes it will be watertight in the face of future privacy legislation. “Because you’re not matching individual IDs, this exceeds the regulatory requirements, in our view,” he says. The probabilistic methodology, Kell adds, is “totally respectful of consumer privacy.”

Though Havas expects Converged will provide a solution to the questions posed by cookie deprecation, Kell also hopes it will prompt higher overall investments from advertisers.

“Clients are under so much pressure to prove the value of their media investment,” she says. “Without cookies, measurement is going to be difficult. We want to get clients looking at the value of media based on results, so we need to have new ways of measuring the impact.”