Google Ad Manager Shifts to Modular, Flexible Publisher Infrastructure Strategy

Full analysis

Google's own director of entertainment and sports partnerships, Brian Jankovsky, stood up at Cannes and said the quiet part out loud: Google Ad Manager won't be sold as all-or-nothing anymore. Publishers can take just the ad server, just the server-side ad stitching (dropping ads into a video stream before it reaches the viewer), or the whole stack. Google, the company that built its ad business on tying its pieces together, now says it'll happily be one Lego brick in someone else's box.

What's actually being decided: Nothing, by you, today. This is a signal, not a fork in the road. The real question for operators is whether the incumbent conceding "modular is the new normal" changes how you write your next RFP, renegotiate your GAM contract, or price your own product against Google. Type 2, easily reversible — nobody has to commit to anything. Forcing function is soft: Q3/Q4 RFP season, plus the drumbeat of Google's ad-tech antitrust remedy in the background.

The Skeptic. Jankovsky said this at Cannes, on the Croisette, between rosé and yacht parties. That's the venue for aspirational statements, not architectural commitments. Google has worn the "open" costume before — header bidding accommodation, open bidding, Privacy Sandbox — and every time the plumbing still favored GAM's own data. The load-bearing claim is that Google will price, support, and instrument a bare ad server at true parity, with no quiet reporting or attribution advantage when you bring your own SSP. That has never held. Plain version: Google saying it'll be "just a pipe" doesn't mean the pipe stops leaking value back to Google. Swap one component out and you'll find the seams.

The Market Analyst. Here's the tell worth reading: incumbents don't preach interoperability when lock-in is working. They preach it when contracts are wobbling and they'd rather write the breakup story themselves than let the DOJ or a defecting publisher write it. This validates the whole independent video-and-ad-infra category — FreeWheel, Roku's SpringServe, Magnite, and the modular-video crowd all sell exactly the unbundling Jankovsky just blessed. Plain version: the biggest player admitting it can't hold the full stack makes everyone else's "pick your pieces" pitch look less risky. Watch whether Magnite or PubMatic name-drop "publisher stack diversification" on the next earnings call. If they do, this quote becomes an industry talking point, not a one-off.

The Operator. The practical win is smaller and more real than the strategy decks suggest. For years, Google's competitors hit the same wall: "You can't partial-in Google, so why bother." That objection just died. Publisher ad-ops leads who never had budget authority — because the choice was all-or-nothing — now get to scope an RFP that says "SSAI only" or "ad server only." That's leverage in a renewal conversation even if you never switch. Plain version: you can now use Google's own words to squeeze a better deal out of Google. What breaks at 90 days: someone actually swaps a component, then discovers the reporting reconciliation between two vendors is a part-time job nobody budgeted for.

The Customer / End User (the large media company). Big publishers have been asking for exactly this — they want to own the parts that touch their audience data and rent the commodity plumbing. Jankovsky's read on his own customers is correct. But wanting control and staffing control are different sports. Owning your ad-serving orchestration means owning the on-call rotation when a stream fails to stitch at 8pm on game night. Plain version: modular means you get the steering wheel and also the repair bill. The publishers who win here are the handful with real engineering benches. Everyone else will "go modular" in the RFP and quietly re-buy full-stack within a year.

The sharpest tensions:

1. Real pivot or retention play? The Market Analyst reads this as Google conceding the moat is eroding. The Skeptic reads it as a retention strategy in open-stack clothing — keep publishers on GAM-the-ad-server so they never fully leave. Both can be true, and that's the trap: the announcement validates competitors while the implementation keeps Google sticky.

2. Optionality vs. inertia. The Operator sees new leverage in every renewal. The Customer persona counters that publishers have resisted unbundling for years because complexity is expensive. The signal is free; acting on it costs headcount most publishers don't have.

What it hinges on: Whether Google supports a stripped-down GAM at genuine parity — same reporting fidelity, same attribution, no preferential data access — when you bring your own SSP or SSAI. If yes, the composable-stack market is real and best-of-breed vendors re-rate up. If no, this is a talking point that helps competitors in the pitch and helps Google in the contract.

Where the council leans: The near-term value is almost entirely as negotiating leverage, not as switching activity. Every operator should walk into their next GAM renewal quoting Jankovsky back at Google. Very few should actually rip out a component this year. Before anyone does, pilot one swap on a low-stakes property and measure the reporting gap — that's where the "quietly irreplaceable" full-stack lives.

Prediction: On their next earnings calls before the end of 2026, at least one of Magnite or PubMatic will explicitly cite publisher stack diversification or unbundling as a tailwind — but no top-20 publisher will publicly announce ripping a core component out of Google Ad Manager in that same window.

Confidence: Medium — the competitive incentive to cite this is strong; the switching cost that stalls actual defection is stronger.

Why: SSP executives are already primed to sell the unbundling thesis, and an incumbent publicly blessing modularity is exactly the third-party validation they quote on calls to explain new logos — that behavior is well-established, so expect it again this cycle. At the same time, actually swapping a GAM component means eating reconciliation, attribution gaps, and on-call ownership that most publishers haven't staffed, which is why years of "openness" signals haven't produced visible defections. The opposite outcome — a marquee publisher loudly leaving GAM's core within months — is the less likely one because these decisions run on multi-year contracts and engineering roadmaps, not Cannes soundbites.

Revisit by 2026-12-31: We're right if a Magnite or PubMatic earnings call cites publisher unbundling/diversification as a growth driver AND no top-20 publisher announces a core GAM component swap. We're wrong if a major publisher publicly dumps a core GAM component this year, or if neither SSP mentions the theme.

The cleanest tell to watch isn't a press release — it's the language in Q3 and Q4 publisher RFPs. If "ad server only" starts appearing as a scoped line item, the pivot is real. If it doesn't, Jankovsky gave a great Cannes quote and not much else.