Future of Marketing Briefing: Horizon Media is building a platform to orchestrate ad tech from a single command center

This Future of Marketing Briefing covers the latest in marketing for Digiday+ members and is distributed over email every Friday at 10 a.m. ET. More from the series →

Agency trading desks spent a decade trying to live down what they were. Horizon Media thinks it’s finally figured out what they should become. 

It’s building what it describes internally as an “orchestration and intelligence” layer that sits above ad tech rather than alongside it. The idea: plug every relevant media marketplace partner – demand-side platforms, supply-side platforms, ad verification vendors, data partners, measurement providers and creative tech – into a central command center of sorts. That command centre, dubbed “Horizon OS” – shorthand for Open Sphere – has already sent RFIs to more than 200 potential partners, which have been whittled down to a pilot cohort now in a green-lit integration phase. 

“The whole idea is we are canvassing the entirety of the media marketplace and basically saying, ‘hey, media marketplace partners, you are being invited in to integrate with Horizon’s central brain’, which will become the elevated orchestration and intelligence layer,” said Maikel O’Hanlon, svp and managing director, programmatic at Horizon. 

The pitch will sound familiar. Every major agency holdco has claimed some version of this – routing ad dollars smarter, more transparently and more in clients’ interests. Publicis has Epsilon. GroupM has built out its own supply chain infrastructure. Omnicom has Acxiom. The agency has been promising a unified intelligence layer since before anyone called it that.

Time and again those promises came down to a human decision made upfront: allocate a portion to The Trade Desk, another to Google’s DV360, write the brief, run it and report back. What Horizon is actually arguing is more specific – a system updating those allocations in real-time based on live performance signals across connected programmatic pipes rather than a media plan that’s essentially set in stone before the first impressions clears. 

“I can’t have 15 different logins for my trading team to go in and try to optimize. Even if the logins have LLMs and agents, it’s really difficult for someone to actually make sense of all the plugin points,” said Nav Singh, managing director of programmatic at Horizon.  

In practice that might mean routing a campaign through The Trade Desk’s pipes at one point, Amazon’s at another or directly through supply-side bidders like Clearline – determined by whichever path delivers the best outcome for a given client at a given moment. 

“Right now, if you ask us to do that, it’s like, okay, we’ve got to set up very specific infrastructure for each single one, and you kind of lose sight of a lot of what you’re doing,” said Singh. “You have to have literal conversations across the board, and you need mastery of all those different things.”

The more meaningful distinction, though, is what gets monetized.

The original agency trading desks made their money on inventory – bought in bulk, marked up, frequently without telling clients. That’s the model the Association of National Advertisers transparency report in 2016 torched, and the one the industry never quite finished reckoning with.

Horizon’s version monetizes the decision making itself: a cut only where it can demonstrate added value – proprietary data, custom bidding logic and measurable performance lift. Platform and data fees, O’Hanlon said, are passed through to clients at cost rather than treated as a margin opportunity – a structural choice he acknowledged creates its own friction, since transparency on line items makes them easier to debate. Intelligence is the product. Not the inventory.

What that means in practice, and how it prices the orchestration layer itself, is less defined. What is clear, though, is the market context it’s operating in: buy-side costs on The Trade Desk can run anywhere from 15% to 30% depending on what’s included, Google’s DV360 sits closer to 4%, with Amazon undercutting both – often, according to people familiar with the economics, with comparable outcomes. If Horizon;s intelligence layer is going to command a fee on top of that stack, it will need to show its working,

“We’ve always treated platform fees and data fees as pass-through components, and so there’s a level of transparency that has always been core to how we operate that is different from the way in which other outfits like ours were built to be profit centers,” said O’Hanlon. “Once you make it transparent, you know, then it becomes easier to debate.”

And it’s a debate that’s expected to swell as more agencies look to upend the way they buy media and therefore how they get paid for doing so. For some that will be principal media, for others it’s about building their own tech to either challenge or orchestrate ad tech. Either way, expect the transparency debate to look very different in two years’ time – less about what ad tech vendors charge, more about whether the intelligence layer sitting above them can prove it’s worth paying for at all. 

“Infrastructure like the Horizon OS offers beneficial efficiency and insight across the ad tech stack for the agency, but the issue the agency trade desks ran into was accountability and transparency,” said  Kym Insana, founder and president at AlwaysOn Digital. “If this OS doesn’t address those issues, then does it really solve agency trading desks’ reputation problem? Until clients have full visibility into how partners are selected, what fees are embedded, and how performance is measured independently, ‘orchestration and intelligence’ risks being a new language for the same old arrangement.”

That tension – between openness and commercial sustainability – is one the wider industry is only just beginning to feel. What’s different this time, Singh argued, is that AI makes the status quote increasingly untenable. More connection points, more live data, more verifiable outcomes means less room to hide margin in the middle of the supply chain – and the people currently doing so, he suggested, can see it coming. 

“It’s like a fire that fizzles before it dies out,” he said of the current fee debate. “Everyone’s trying to hold on so deeply to the margins because they’re so desperate for them. They don’t have a better model right now.” 

Numbers to know

$852 billion: How much OpenAI has been valued at, having raised its latest $122 billion in capital.

$1.75 trillion: The target valuation Space X is reportedly after for its IPO.

$4.2 million: The amount Meta must pay in combined compensation and punitive damages over losing the social media addiction case in California.

47%: Percentage of agency execs who said limited platform integration and data unification are marketers’ top barriers preventing scaled personalization

What we’ve covered

Pinterest bets measurement and SMBs will boost performance revenue

Pinterest’s latest reorg brings product, engineering and data science under one leader, Vik Gupta, who recently joined in the newly-created role of vp, gm of monetization. In his view, measurement is central to winning more ad spend and smaller advertisers.

In graphic detail: The long road to accountability for social media platforms

The verdicts in the U.S. California and New Mexico trials against Meta and YouTube signal a fundamental shift – that platforms can now be judged on how they’re built and designed, not just the content they host. Digiday charted what led to this point.

‘Predictability has become a luxury’: As the Iran war drags on, ad markets are starting to sweat

Any hopes that this would be a short war, and slowly fading away. Now it’s about figuring out how to minimise as much economic damage as possible.

AI influencer discovery tools are changing how agencies cast creators

Agencies are using AI to automate far more of the influencer marketing workflow than they used to.

What we’re reading

Can you have child safety and Section 230, too?

Platformer dives into the verdicts from both the California (against Meta and YouTube) and New Mexico (against Meta), how section 230 has protected platforms until now, and how platforms will likely require changing core product features in the future.

TikTok Abandons Plans to Add Second Data Center in Ireland

TikTok has decided to drop its plans to build a second data centre in Ireland to house its European user data, and is instead focusing on other projects which will store European user data in the region, per Bloomberg.

Irenic Capital Discloses Snapchat Stake, Seeks Strategy Changes to Boost Valuation

Activist investor Irenic Capital has published its letter to Snap CEO Evan Spiegel, along with its 70-page deck on why it believes the company should be worth $35 billion by now, but has instead underperformed, while providing suggestive steps to make it profitable, according to the Wall Street Journal.

Amazon’s AI Chat Ads Yield Data but Few Sales

According to The Information, shoppers are viewing and clicking on Amazon’s chatbot ads a lot less than other types of Amazon ads, meaning they can’t figure out how well they drive sales. But on the bright side, they are cheaper.

More in Media Buying

Transparency brought down The Trade Desk’s Publicis deal but it may not be enough to win its other clients

Nexxen’s launching new AI features. Amazon’s broaching streaming partnerships. And Viant, Blockboard and Stackadapt are setting out their stalls.

Spotify’s ad exchange grew its programmatic ad base, but buyers want more

The number of advertisers using the music and podcast streamer’s programmatic marketplace has risen 222% since last year.

Ad Tech Briefing: Criteo makes self-serve push with GO, targeting SMB growth

Criteo is bringing itself into more direct competition with Amazon and Google.