While Google’s cookie clock ticks, one sports publisher is betting on AI—not cookies—to keep 15 million fans scrolling and shopping.
On3, the fast-growing network of 70+ college-team fan pages, has quietly plugged 20 of its sites into an AI-driven recommendation engine built by startup Mula. The goal? Squeeze more revenue out of every session without chasing the next social algorithm change or pouring money into more headcount.
The early returns are hard to ignore. On3’s 15 million unique monthly visitors and 3 million email subscribers now see a dynamic feed that mixes editorial, e-commerce, and native ad units—each ranked in real time to maximize revenue per session (RPS). In practice, that means a Notre Dame fan reading about spring roster cuts may be nudged toward a $2,500 autographed mini-helmet sold via Fanatics, or a related story about the incoming recruiting class surfaced only if the model predicts another pageview—and another ad impression.
“We’re always looking for ways to drive users to stay,” says Brandon O’Neal, On3’s Chief Revenue Officer. “Social media has done such a good job with that ‘scarcity brain’ mentality. We needed a way to replicate it on-site.”
Why RPS is the new pageview
Traditional sports publishing still lives and dies by vanity metrics: monthly uniques, Facebook reach, search ranking. On3’s leadership decided those yardsticks are broken. After watching Google referrals slide and social bounce rates climb, the company asked a simpler question: How do we make more money from the users already here?
Enter Mula, an “agentic publisher monetization platform” that ingests only first-party behavioral signals—clicks, scroll depth, cart adds—and predicts what a reader is most likely to click or buy next. Every three seconds the model re-scores the inventory of stories, merchandise, and sponsored links, then shuffles the on-page modules accordingly.
“RPS is the god metric,” says Jason White, Mula’s Co-Founder & CEO. “If we can’t raise it by double digits, we don’t deploy.”
Across the initial tranche of On3 properties, the uplift has hovered between 18 % and 34 %, according to White, with no additional ad ops staff or custom dev work.
A cookie-less testbed bigger pubs should envy
Because Mula’s models run exclusively on first-party event streams, On3’s recommendation engine is already cookie-less—and by extension Privacy-Sandbox-ready. That gives endemic advertisers like Fanatics and RevContent a rare sandbox to test personalized, high-ticket placements without leaning on third-party IDs or the crumbling cookie.
“The way we look at it is if we’re providing value to our users, it’s something worth trying,” O’Neal says. “If the AI surfaces a helmet and a fan buys it, everybody wins.”
The setup also shields On3 from the SSP roll-up mania sweeping ad tech. Instead of bulking up through M&A and handing more margin to Google, On3 is effectively turning itself into a self-contained DSP for sports-obsessed budgets—an approach that preserves data and keeps buy-side fees in-house.
Commerce beats CPM—no video required
Media analysts love to tout connected-TV CPMs north of $30, but On3’s memorabilia sales hint at fatter margins hiding in plain sight. A single $2,500 helmet nets the publisher a low-double-digit affiliate cut—the equivalent of a $300-plus CPM after just a handful of impressions. No studio, no pre-roll, no audience extension required.
RevContent CEO Richard Marques sees the pattern spreading.
“Commerce can be a high-margin, diversified revenue channel for publishers, but it’s a full-time gig,” he notes. “AI ranking removes the operational headache.”
Expect other niche publishers—think golf, fishing, crypto—to flip their editorial calendars into “shoppable liveblogs” using the same ranking logic. If a $50 driver grip or a $400 putter converts, the effective CPM dwarfs what display or even video can deliver.
What happens next
On3 plans to roll Mula’s feed across the remaining 50+ team pages before football season kicks off, layering in video recommendations from partner EX.CO and more granular merch SKUs from Fanatics. White says the platform will soon optimize push-notification copy and newsletter module order, chasing RPS gains in channels most publishers still treat as flat, low-yield inventory.
For an industry stuck trading the same open-market impressions, the takeaway is clear: own the audience, own the data, let AI figure out the rest. If On3 can turn die-hard fans into high-roller collectors without extra staff, the rest of publishing might soon trade pageviews for shopping carts—and never look back.
💡 Deep Dive: Don’t miss our Ultimate Industry Guide for advanced strategies.
