Why Performance Max advice keeps changing: the strategy debate through 2025 and 2026

Written by ChatGPT (AI) · edited by Nathan Pankowsky.

Give Performance Max every asset it asks for. Keep it feed-only. Consolidate your campaigns. Separate your margins. Exclude your brand. Be careful excluding your brand.

Anyone researching PMax strategy will encounter advice that points in different directions. Some disagreements reflect different businesses and objectives. Others reflect a product that has changed substantially while the discussion continued around it.

A recommendation made before channel reporting, campaign-level negative keywords, or newer allocation controls needs to be read in that context. It might still be useful. Its original justification may also have weakened.

This article follows several practitioner discussions alongside Google's product announcements. The aim is to explain what each strategy was trying to solve, what changed, and which difficulties remain. It is a selected history of the debate, with early-2025 context for changes that shaped the past year.

How the available decisions changed

Google's announcements show a progression from more ways to guide PMax toward more ways to inspect and constrain its delivery.

The distinction between a feature announcement, an experiment, and availability in a particular account matters throughout this timeline.

Debate 1: feed-only PMax, full creative, or Standard Shopping?

One longstanding strategy has been to build PMax around the Merchant Center feed while withholding manually supplied creative. The attraction is understandable: an advertiser wants product sales and is wary of sending budget into less familiar formats.

In Optmyzr's PPC in 2025 discussion, Andrew Lolk described his earlier support for feed-only PMax, then questioned why an advertiser wanting that approach would choose it over Standard Shopping. His argument emphasized control over the type of inventory being purchased. [8]

That is a useful example of a practitioner revisiting a recommendation rather than treating it as permanent doctrine.

A different argument favors supplying creative so PMax can pursue customers across more formats. But a complete asset set is an input, not proof of commercial improvement. Optmyzr's analysis of its 2025 updates and campaign study questioned any simple relationship between asset completeness and superior performance. Its observational findings should not be read as a controlled test of whether adding assets causes better or worse results. [9]

Google now documents experiments comparing feed-only asset groups with groups supplied with additional text, images, and video. [10]

Our interpretation: the strongest version of this discussion asks what incremental value additional creative produces for a particular business. Feed-only should also not be treated as a guarantee that every impression is a Shopping ad. Inspect actual delivery and compare alternatives against profitable outcomes.

The newer tools make this question more testable. They do not establish a universal winner.

Debate 2: exclude branded searches—or understand their role first?

Brand traffic creates an awkward reporting problem. Someone searching for your company may already intend to buy. A campaign can receive conversion credit without establishing how much additional demand it created.

That supports the case for excluding brand searches from PMax and managing them separately. It also explains why a high blended ROAS can leave an advertiser uncertain about prospecting efficiency.

Mike Ryan's April 22, 2025 PMax Q&A adds a practical complication. Responding to a campaign that flatlined after a brand exclusion, he advised checking how much conversion volume had been removed, allowing for disruption, and reconsidering whether the target remained appropriate for nonbrand traffic. He cautioned against applying a common recommendation without understanding its scale and tradeoffs. [11]

By October 2026, Menachem Ani was also discussing brand cannibalization as a condition to investigate within a broader campaign-structure decision. His framework resisted both excessive consolidation and excessive isolation. [12]

Our interpretation: brand exclusions can clarify a measurement question, but they also change the traffic the campaign can pursue. A fall in campaign ROAS after an exclusion needs to be assessed alongside total business outcomes and the economics of the remaining traffic.

There is a further reason to check older discussions carefully. Ryan's 2025 article describes negative keywords as applying more broadly across inventory than Google's current documentation supports. Current Google guidance limits PMax negative keywords to Search and Shopping inventory. We cite the article for its strategic caution, not that implementation claim. [11][13]

Debate 3: consolidate for learning or separate for economics?

Consolidation gives each campaign more activity to work with. Separation gives the advertiser distinct budgets and targets. Both solve real problems.

In the Optmyzr discussion, Lolk strongly criticized splitting campaigns that effectively shared the same commercial target. He also allowed for separation where product categories needed meaningfully different ROAS targets. [8]

Ani's October 12, 2026 framework puts the tension into account-specific terms: conversion volume, product margins, feed readiness, brand traffic, and channel performance should influence the structure. A structure suitable today may need revisiting as those conditions change. [12]

These positions have substantial common ground. Campaign count should reflect a decision the business needs to make. A separate budget for a different market or margin profile has a clearer purpose than separation solely because a catalog has another category.

Our interpretation: more reporting can reduce the need to create campaigns merely to obtain visibility. It cannot eliminate a business requirement for separate budgets or targets. The practical difficulty is giving each commercial boundary enough activity to evaluate and optimize it.

Debate 4: does visibility justify channel restrictions?

Channel reporting changed the discussion by making delivery easier to inspect. It also introduced a temptation: rank channels by attributed CPA or ROAS, then reduce whichever looks worst.

Google's April 2025 announcement explicitly cautioned that average channel returns do not capture the full picture of marginal optimization or customer journeys. [2]

That warning deserves consideration. A channel might help introduce customers who later convert elsewhere. Equally, an advertiser needs a way to investigate spending that consistently fails to produce valuable outcomes.

The familiar slider concept addresses part of that tension: express a preference and give automation bounded room to explore. But different iterations of channel controls must be distinguished.

The October 2026 official rollout documents High and Low priorities plus optional budget allocations for lower-priority inventory. In PMax, Search, YouTube, Maps, Gmail, and Discover are grouped as Google channels; Search partners and the Google Display Network can receive separate priorities. The controls therefore do not provide an independent budget slider for every Google-owned surface. [6][7]

Google also says prioritization has no effect when a campaign is not budget constrained, while allocations still apply. Allocations operate on a best-effort basis. [7]

Our interpretation: this creates a useful middle ground between unrestricted allocation and excluding inventory altogether. It also shifts the advertiser's task toward deciding which evidence warrants a restriction. A low attributed return over a short period is weaker evidence than a sustained pattern supported by qualified outcomes and a suitable test.

Debate 5: what are we teaching it to chase?

A cheaper form submission is lovely. Sales would also like someone to call back.

That is the catch behind many PMax success stories: a submission, a qualified opportunity, and a sale are three different reasons to celebrate. Google recommends feeding downstream lead outcomes back into optimization, while balancing quality against having enough data to learn from. [14]

The practical snag? Your CRM may still be catching up. Give leads time to qualify, check that outcomes are recorded consistently, and judge the campaign by the customers it brings. Otherwise, you can spend a week adjusting sliders while the real problem sits in the conversion column.

Before reaching for another slider

When the numbers look odd, it helps to follow the visit a little further. What arrived? What did it do? Did anything useful happen afterwards?

That is where Button’s Ad Protect fits: investigate browser and journey evidence alongside campaign results, then decide whether to change your feedback settings.

Keep a little curiosity in the process. A brief visit can be human. A convincing browser can still produce a terrible lead. And website evidence cannot magically name the publisher behind every click; Google's placement report has limits too. [15]

The useful outcome is a clearer next move: review a landing page, investigate a traffic pattern, or test a boundary the new controls make possible.

Best practice, with a date stamp

PMax advice has a shelf life. Before borrowing someone else's winning setup, check when they used it, what they were selling, and what counted as a win.

The newer controls are welcome progress: more ways to investigate, more room to experiment, and fewer reasons to treat every decision as all-or-nothing.

Pick one question. Make a change you can evaluate. Give the results time to arrive.

And if someone promises the one perfect PMax setup, check the publication date before rearranging your account.

---

References

Dates above refer to announcements or publication dates, not universal account availability. Practitioner positions are attributed views, not consensus. Paragraphs marked “Our interpretation” are Button editorial analysis. No client results or causal performance claims are presented.

1. Google, January 23, 2025: Kick off 2025 with new Performance Max features.
2. Google, April 30, 2025: Channel performance and more reporting coming to Performance Max.
3. Google, August 7, 2025: Unlock more visibility and control in Performance Max.
4. Google Ads Developer Blog, January 29, 2026: Introducing channel-level reporting for Performance Max campaigns.
5. Google: New Performance Max steering and reporting updates coming in 2026.
6. Google, October 14, 2026: New channel prioritization and budget controls.
7. Google Ads Help: About channel prioritization and budget allocation.
8. Optmyzr, practitioner panel discussion: PPC in 2025: What to stop, start and double down on. See Andrew Lolk's discussion of feed-only campaigns, Standard Shopping, and consolidation.
9. Optmyzr: Google listened: 5 PMax fixes that solved the gaps we found in 24,702 campaigns. Observational analysis; asset comparisons do not establish causality.
10. Google Ads Help: About Performance Max optimization experiments: Asset testing.
11. Mike Ryan, April 22, 2025: Performance Max Q&A: Real answers to your biggest PMax headaches. Read implementation claims alongside current documentation, particularly inventory scope for negative keywords.
12. Menachem Ani, October 12, 2026: Consolidate or segment? A Performance Max decision framework. First-person practitioner commentary.
13. Google Ads Help: How to use brand suitability features in Performance Max.
14. Google Ads Help: Answering your top questions about Performance Max.
15. Google Ads Help: Find or create placement reports for your Performance Max campaigns.