How Tripadvisor and PMG are breaking down the brand-agency divide with an ‘in-residence model’

Full analysis

PMG puts its people inside Tripadvisor's building. Badges, desks next to the brand team, visibility into product and finance. Full access to the brief and beyond it. Carly Carson of PMG calls it an "in-residence" model. Kristen Tormey of Tripadvisor/Viator runs it alongside two other agencies and drags media platform reps into quarterly in-person sessions where creative, analytics, and product all have to show up. That's the whole story. No numbers, no deal, no capability drop.

So let's be clear about what's being decided here, because it's not really a decision at all. It's a positioning signal. The question worth chewing on: is "we'll move into your office" a durable answer to the pressure on agency value as AI eats the mechanical parts of media buying? Type 2, easily reversible for any single account. Structurally, the trend it points at is not.

The Market Analyst Watch what PMG is selling, because it isn't media execution. As automated buying gets commoditized, the retainer for "we push the buttons" compresses. So the independents differentiate on proximity: embed, get the badge, become essential to the client's org chart so you're harder to rip out. For a generalist: when the work a middleman does gets automated, the middleman sells being close to you instead. This is the independent-agency answer to the holdcos, who sell scale and data. PMG can't out-scale Publicis, so it out-integrates them. Expect more of this framing at every independent shop pitching against a holdco in 2026.

The Skeptic The case against isn't hard to make. "We want desks in your building" is a lovely story that survives exactly as long as the client champion does. Embedding is expensive labor dressed as intimacy, and it doesn't scale past your biggest accounts. Tormey calling Tripadvisor "a bit of a challenger brand" tells you the register here is vibes, not rigor. A publicly traded platform with hundreds of millions of users is not a challenger. When a partnership story leads with culture and Demi Lovato community management, ask where the incremental return is. They mention IROAS targets, then don't share one number. That silence is the review.

The Operator The genuinely useful bit is Tormey's rule that platform sales reps can't run siloed conversations. She makes The Trade Desk and TikTok bring creative, analytics, and product to the table, not just an account rep with a roadmap deck. Any brand or agency lead can copy that Tuesday morning. It costs nothing and it changes what you get out of a partner QBR. The in-residence model is the harder lift: embedded staff means somebody's carrying that headcount, and the second the account softens, those badges are the first cost a CFO circles. Integration that deep is a retention moat when times are good and a liability when they aren't.

The CFO Somebody pays for the desks. In-residence is more bodies, more travel to New York for quarterlies, more coordination overhead across three agencies convening on one plan. The pitch is that this proximity produces speed and better outcomes, and maybe it does. Untested here, in the numbers they shared. And it'll stay that way, because the episode offers IROAS and brand lift as words, not figures. For the agency, this model trades margin for stickiness: you spend more to serve the account, betting the client won't leave a partner who's woven into product and finance. Fine bet on your top accounts. Ruinous if you promise it to everyone.

The tensions. The Analyst says proximity is the smart structural play for an independent that can't win on scale. The CFO says proximity is margin you're giving away, and it only pencils on your biggest logos. Both are right, which means the real question is how many accounts can actually carry an embedded team. The Skeptic and the Operator split on the same fact: the shared-data-foundation-before-AI point. The Operator sees a real gate every buyer now applies. The Skeptic hears table stakes wrapped in future tense.

What it hinges on. One belief: that as AI automates buying, clients will pay a premium for an agency that's operationally fused to them rather than one that's merely good at execution. If that's true, in-residence is a moat and the holdcos have a problem, because scale and proximity are hard to sell in the same breath. If it's false, it's expensive theater that evaporates with a CMO change. The council leans toward the trend being real and the specific model being oversold. Carson's data-foundation line says it plainly: everyone now sells "get your data clean, then we'll do AI," and that gate is becoming the standard opening question in every optimization pitch.

Prediction: Through the 2027 new-business cycle, at least two more major independent agencies (the tier of PMG, Horizon, Mediahub) will publicly market an embedded or "in-residence" client model as a named differentiator against holding-company pitches, framing physical/operational integration as their answer to AI-driven buying automation.

Confidence: Medium The structural pressure is real. Timing and who moves first are not.

Why: Automated buying is commoditizing the mechanical media work that justified agency retainers, and independents can't beat Publicis, WPP, or Omnicom on scale or data assets, so they need a different axis to compete on. Proximity is that axis, and PMG naming and marketing "in-residence" here is the first clear packaging of it. The pattern in agency-land is that a differentiator, once it wins pitches, gets copied and named within a year. The opposite outcome, everyone staying quiet on integration, is unlikely precisely because the AI-eats-buying anxiety is loud and every independent needs a story that answers it.

Revisit by 2027-06-30: We're right if two or more independent agencies of comparable scale publicly pitch a named embedded/in-residence model as a holdco counter by mid-2027. We're wrong if in-residence stays a PMG-specific talking point with no named imitators.

The one thing worth stealing regardless of where you land: make your platform partners bring the whole team to the room. That's free, and it's the only concrete operating change in the episode.