PMax, Amazon Ads, and the Slow Return of Control

The two biggest walled gardens in digital advertising are, quietly and reluctantly, handing some control back. Not because they’ve had a change of heart — because advertisers with real budget weight pushed hard enough, long enough, with enough evidence that the machines were working against them.

This week offered two separate dispatches from that ongoing negotiation. Both are worth reading together.

Google Blinks on PMax — Sort Of

Digiday reports that Google has begun offering media buyers more visibility and override capability within Performance Max — the campaign type that, since its 2021 rollout, has functioned less like an ad product and more like a strategic ambiguity generator.

The concessions are incremental: improved search term reporting, expanded brand exclusion controls, and early access to asset-group-level performance breakdowns for select advertisers. Not a transparency revolution. But for anyone who has spent the last three years trying to explain PMax performance to a CFO using a handful of aggregated metrics, it’s meaningful movement.

The strategic read here isn’t that Google is becoming more advertiser-friendly. It’s that the criticism finally became commercially uncomfortable. Large retail and e-commerce advertisers — many running significant SEA operations across Google Shopping and Discovery — had started routing budget toward more controllable environments. When budget follows the complaint, platforms listen. The practical implication for media buyers: document your control gaps with spend data, not frustration. Platforms respond to revenue risk, not feedback forms.

Amazon’s Sellers Are Done Being Polite

AdExchanger’s reporting from inside a meeting of Amazon’s million-dollar seller cohort paints a sharper picture. These aren’t fringe operators — they’re high-volume merchants whose ad spend funds a meaningful slice of Amazon’s $56 billion+ advertising revenue. And they’re furious.

The core grievance: Amazon has made changes that shift where seller ads run and compress the margin math that made Amazon advertising defensible in the first place. Sponsored Products increasingly appearing off-platform — in third-party apps and sites via Amazon DSP — means sellers are paying for impressions with lower purchase intent and weaker attribution, without explicit opt-in.

This matters well beyond the US. Across Southeast Asia, Lazada and Shopee have built ad ecosystems that closely mirror Amazon’s model — sponsored placements, off-platform retargeting, and performance metrics that the seller has limited ability to interrogate. The Amazon seller revolt is a preview of conversations that will happen on regional platforms within 18 to 24 months, as their ad products mature and seller sophistication catches up.

---

The Pattern Underneath Both Stories

Strip away the platform-specific details and the same dynamic is operating in both cases: automation and scale were sold as advertiser benefits, but they also happened to route more decisions — and more margin — toward the platform. Advertisers who accepted that trade-off during growth years are now reassessing it during a period of tighter scrutiny on media efficiency.

The buyers making headway aren’t the ones issuing position statements. They’re the ones arriving at platform conversations with clean measurement frameworks that isolate what the black-box product actually contributed versus what would have converted anyway. Incrementality testing isn’t a new concept, but it’s become the primary leverage tool in these negotiations. If you can show Google or Amazon that a portion of their attributed conversions are organic cannibalization, you have a number they have to respond to.

For teams running programmatic across Southeast Asia, the regional nuance is this: platform dependency here runs deeper than in Western markets. Google and Meta dominate upper-funnel. Shopee and Lazada own the bottom. There is less inventory diversity to route budget toward when you want to punish a platform. That makes the measurement argument even more important — it’s often the only lever available.

What This Means for Your Next Planning Cycle

Three things worth building into your H2 planning if they aren’t already:

First, run a PMax containment audit before Google’s new controls are fully available in your market. Map which campaigns have brand exclusions properly configured and which are running without them — the new reporting will expose gaps you didn’t know existed.

Second, for any team running on Amazon or its regional equivalents, pull a placement-level breakdown and calculate ROAS by on-platform versus off-platform inventory. If you’re subsidising third-party network impressions at the same bid as high-intent product pages, you’re leaving margin on the table that belongs in your pocket, not the platform’s.

Third, make incrementality testing a budget line, not an afterthought. A geo-holdout or intent-matched control group study running alongside your largest automation-driven campaigns will give you the evidence needed to negotiate — whether that’s with Google, Shopee, or your own leadership asking why CAC is drifting.

The platforms are not becoming your partners. They’re becoming slightly more negotiable. That’s a different thing, and it requires a different posture.

What does it look like when Southeast Asian brands have enough measurement sophistication to actually use these new controls — or are most teams still two capability gaps away from making them matter?

---

At grzzly, this is the kind of problem we spend a lot of time inside — building the measurement infrastructure and bid strategy frameworks that turn platform concessions into actual efficiency gains for brands across Southeast Asia. If your team is navigating PMax opacity, marketplace ad complexity, or both at once, we’re worth a conversation. Let’s talk