Schibsted Uses Custom AI Algorithms to Win Ad Budgets Back From Meta

Nearly two thirds of advertisers — 63.9% — have shifted mobile budgets away from walled gardens and toward independent partners in the past year, according to a July survey of 166 brands and agencies by eMarketer and Datum Intelligence. Only 1.2% moved budgets in the opposite direction. Norwegian media group Schibsted intends to capture its share of that reversal, and it is betting on custom ad optimization algorithms built by London-based Nodals AI to do it.

The shift is real but modest. eMarketer still projects that Meta, Google and Amazon will attract 62.3% of worldwide digital ad spend this year, with Meta topping Google in ad revenue for the first time. Publishers fighting that concentration need more than reach.

Why reach stopped being enough

Schibsted owns VG, Norway's largest digital news publisher with about 2.4 million daily visitors, and Aftonbladet, one of Sweden's largest, claiming 3.6 million daily visitors. Last year it added Sweden's TV4 and Finland's MTV Oy — two of the biggest TV broadcasters in those countries — in an acquisition from telecom group Telia. Combined with its local newspaper network, the company reaches roughly 50% of audiences in its markets, said Christer Ljones, director of ad data capabilities at Schibsted Marketing Services.

For perspective: VG and Aftonbladet each reach more readers than the Los Angeles Times, which draws a combined weekly online and print audience of 4.4 million — about 11% of California's population.

VG and Aftonbladet earn most of their money from advertising; local titles lean on subscriptions. Online display and branded content split ad revenue roughly evenly, and direct sales account for about two thirds of the total, with programmatic making up the rest.

Yet reach alone has stopped winning performance-minded brands. Ljones said walled gardens have won advertisers over with AI-powered optimization tools such as Meta's Advantage+, and open-web publishers have struggled to match that ad tech.

What changed in the sales pitch?

Nodals Founder and CEO Aly Nurmohamed said direct sales grew harder for publishers as Meta reshaped advertiser expectations. "Advertisers have been trained by Meta to just care about 'I put money in and I see how much money comes out,'" he said. Brands were telling Schibsted: "I don't want to buy on a CPM, and you're really expensive relative to Meta. Here's what I'm doing with Meta. Can you replicate it?"

The Nodals approach feeds a publisher's first-party data into its own models — logged-in audience data, scroll depth, click frequency on content and ads — to predict who is likely to click and buy from a particular advertiser. Because the publisher owns both the algorithm and the data, Nurmohamed said the system should sidestep Apple's crackdown on third-party audience data vendors and ID graphs.

Schibsted's structural advantage helps. Its owned-and-operated properties keep a direct audience relationship, and search drives only a single-digit percentage of traffic on its largest sites, insulating it from AI search traffic loss.

How is Schibsted pricing the new products?

Schibsted first tested Nodals' tech last summer and has since folded it into direct sales, Ljones said. It is building a new product line of direct-sold native ad formats with AI optimization built in, priced on a cost-per-click basis rather than CPM. Early adopters include traditional advertisers such as grocery chains in Norway and Sweden seeking a mix of performance and brand awareness.

Ljones conceded Schibsted "won't be necessarily as good as Meta on average," but argued that publishers "don't need to beat them on the average. You just need to beat them with meaningful scale at the margin." The pitch targets advertisers seeing diminishing returns on Meta: "We get feedback from advertisers that they've seen their Meta ROI, their conversions, flatten out, or maybe even decline slightly," he said. The working offer: "Let's take 10% of whatever you spend on Meta, and you can probably keep most of your conversions and get more [from us]."

The legal front

Schibsted has also treated Meta as a competitor, not a partner, in court. It joined 180 other publishers and the Norwegian Media Businesses' Association in suing Meta at the end of October over a terms-of-service update that would have required publishers to share data for training Meta's AI models. Within two days, the Oslo District Court froze the update in Norway pending a legality review — a decision that likely contributed to Meta pausing the TOS change across all regions earlier this month.

"We're in an early phase, but we're seeing the right results," Ljones said, describing the goal as taking new budgets from existing clients, defending current spend and winning back clients Schibsted had lost. If Meta conversions keep flattening, that margin is where publisher AI tools will keep competing.