The move reflects a growing push by publishers to lean deeper into first-party data strategies and gain a foothold in the fast-expanding retail media space, where advertisers are eager for better targeting outside of the walled gardens. For Ocado the partnership provides valuable access to upper and mid-funnel data to inform its existing customer data. 

The partnership uses Permutive’s clean room infrastructure to match anonymized publisher audience segments with retail shopper segments. The publishers are working individually via Permutive’s platform, rather than collectively sharing data. 

If, for example, an individual buys a chocolate bar on Ocado, one of the publishers – let’s say Immediate Media with its Good Food title – could potentially match data that shows that person or someone like them read an article about baking cakes for their daughter three weeks ago. “Those signals are very rich across the open internet, but retailers haven’t had the chance to partner with publishers to identify those signals on the path to purchase and then deliver ads against them, “ said Permutive CEO Joe Root. “When you combine those two sets, the conversion data and the signals and the intent signals which drove it, then all of a sudden you can build this full funnel proposition.”

With advertiser budgets under more scrutiny than ever, there’s growing pressure for outcome-based solutions – and to meet buyer demand for more efficiency, publishers are leaning into predictive targeting models, like Permutive’s, as post-cookie audience strategies become more of a default. The models can then be incorporated into PMP and programmatic guaranteed buys. 

“We know every pound is being scrutinized by advertisers at the moment, Charlie Celino, director of commercial services for News UK told Digiday. “You only have to look at the geopolitical issues and the current economic climate, as well as the fact that we know we have to work harder, and advertisers and CMOs are having to work harder for that cash,” he added. 

Meanwhile, the ongoing discomfort publishers feel generally at the rise of Google AI Overviews and AI platforms eroding referral traffic makes deepening relationships with audiences more vital than ever for publishers. Until now, there has been a gap between the rich upper funnel audience data publishers have, like what pages and content readers engage with, and how frequently, and the lower-funnel retail purchase data that online grocery commerce platforms like Ocado have in droves. 

“We continue to move into this world in which traffic is eroding for publishers,” said Matthew Rance, head of commercial data and analytics for Immediate Media. “The web is under threat, it’s destabilizing and [it’s] certainly not in growth anymore. And the only thing you can really bank on now is a publisher is the direct relationship with your customer.”

As addressability on the open web erodes, publishers are pushing to improve their predictive models to help advertisers reach consumers without relying on identifiers. “We know that cookies are only supporting approximately 30% to 35% of web traffic and nothing in Safari,” said Cian Jordan, head of data monetization for The Independent. “So if specific brands [advertisers] want to find consumers who they believe are bottom of the funnel as part of their target audience, we need new ways of identifying them. And so this is one such initiative,” he said. 

The publishers and Ocado have spent months selecting and inputting their data segments into Permutive. Now they’re getting ready to pitch the new capabilities to advertisers. 

Advertisers will get to use the additional data insight to inform and tailor their campaigns across all News UK properties, which include the Times of London, The Sunday Times and The Sun newspapers which have just under 35 million combined scale, according to Ipsos data; The Independent’s main site, along with the various magazine titles of Future and Immediate Media, which has a combined scale of 21 million monthly readers across its titles including Good Food, according to the publisher. 

“We’re hoping that this will result in commercial benefit for us, so that we’ll be able to command a higher CPM for this advertising,” added Jordan.

By partnering with Ocado, publishers and the online grocery marketplace can better bridge the gap between upper funnel media and lower funnel retail conversions.

“Retail is being forced to go full funnel, said Permutive’s Root. “If you’re Amazon, you own home video and Twitch, and it’s quite easy to go full funnel. If you’re a retailer who isn’t Amazon, going full funnel means you need to go and build partnerships with publishers.” This partnership will help scale Ocado’s data in the market, he added. 

Ben O’Mahoney, adtech and data partnerships lead at Ocado Ads, said combining its real-time customer purchase insights with the publishers’ deep understanding of their audiences, securely via the Permutive platform, will provide advertisers with “unparalleled opportunities to reach high-intent consumers with relevant messages.”

Clean rooms have gotten a much-needed overhaul 

Clean rooms gained traction as a way for advertisers and publishers to combine first-party data without violating privacy regulations, and to prepare for third-party cookie deprecation, which Google initiated in 2020. But as adoption grew, so did skepticism around their ability to scale, for being overly complex and failing to deliver the outcomes CMOs expected. 

Celino, Rance and Jordan all believe clean rooms have had a much-needed makeover. 

“There was this obsession [before] with the first approaches to clean rooms and [data] matching with IDs and it created a lot of issues,” said Rance. “Publishers maybe felt stifled, or that they couldn’t really play in this space, because they didn’t have 100% logged-in users all of the time.” 

The strength of the Ocado-publishers relationship is that it matches data that has been gained through years of developing deep relationships with readers, with years’ worth of insights gained from an established database of Ocado’s daily customer shopping habits, said Rance. 

“Even if those kinds of initial matches are across certain segments and are very small, the real power of it is that you’re then using that highest quality kind of seed match to then find people who look extremely similar. There’s no threat to signal erosion or signal loss there,” he said.

Root stressed that clean rooms have evolved from the earlier deterministic iterations to more predictive ones, where instead of sharing personally identifiable data, a privacy-safe model is trained on cohort and contextual signals and then shipped to the publisher. This new approach means there is no need to share personal IDs or information. 

“We’ve really gone from this world of deterministic clean rooms where you push IDs out, to a world of predictive clean rooms where you push a model out instead,” said Root. “And what the model does is it takes a load of inputs, which are the publisher signals, it takes cohorts and contextual signals, and then makes a prediction. But those cohort and contextual signals, they’re not personally identifiable information.”