Google Ads · Video breakdown

PMax vs Search is the wrong fight in 2026

Google put an AI layer on both campaign types, turned it on by default, and justified it with a number that leaves out retail. Here is what independent data on stores shows, and the one decision that still matters.

Watch the breakdownRun the free Google Ads audit
The short answer

Stop picking a winner. Decide what PMax is allowed to eat.

Performance Max and Search now do different jobs, and Google is steering both toward the same AI layer. The decision that still has a right answer is whether PMax is allowed to take your brand traffic. Every branded conversion it absorbs was a customer who already knew your name and was already coming, and it lands in the return column looking like acquisition.

Exclude your brand terms from PMax and every number you look at afterwards gets more honest. The rest of this page is the evidence, and the four things I would do this quarter.

Watch

The 6-minute breakdown

Every number in the video is published and sourced below. No client accounts appear on screen.

Chapters

0:00The fight is over, and neither side won
0:33Twenty seconds on who is talking
0:51PMax got windows, Search got a black box
1:55The 14% claim, and the footnote that excludes retail
2:54250 retail campaigns: revenue +13%, CPA +16%
4:02The real split is brand traffic, not campaign type
5:04Four things to do this quarter

Video page with chapters · Watch on YouTube · More breakdowns on the channel

The numbers

The number Google leads with, and the one measured on stores

14%

more conversions at a similar CPA: Google’s own AI Max claim, which leaves out retail

+13%

median revenue across 250+ retail campaigns with AI Max on

+16%

median cost per acquisition in the same campaigns

+42% to -35%

the spread of outcomes. The analyst’s summary: essentially a coin toss

Read the middle two together. Revenue up 13% and cost per acquisition up 16% means you bought more volume and paid more than proportionally for it. On thin margins that is not a win. It is a bigger version of the same problem. The retail figures are Mike Ryan’s analysis at Smarter Ecommerce, quoted as his finding.

What changed

The two campaign types moved toward each other

PMax

The black box got windows

  • Channel performance reporting, so you can see how budget splits across Search, Shopping, Display, YouTube and the rest.
  • Search term visibility.
  • Campaign-level negative keywords, raised to 10,000.
  • Search themes, doubled.
Search

Search got a black box

  • AI Max: expanded matching, generated ad copy and its own choice of landing page.
  • Campaigns being auto-upgraded into AI Max by default.
  • Dynamic Search Ads being deprecated and folded into AI Max, confirmed by Google’s ads liaison with no timeline.

PMax became more controllable and Search became less. They are converging, and they are meeting on Google’s side of the room. The question stopped being which campaign type to run. It is now which parts of your account you are still allowed to steer.

Different jobs

Benchmarks say division of labor, not a leaderboard

Campaign typeAvg. CPCConversion rateReturnIts job
Search$1.422.8%3.4xExplicit intent, expensive and worth it
Performance Max$0.821.9%4.8xDemand harvesting with a wide net and a cheap click
Shopping$0.681.4%5.1xYour catalog does the selling

Ecommerce averages from CorePPC’s 2026 benchmarks. A higher reported return on PMax partly reflects what it is allowed to serve, which is why the brand-term decision below matters more than the averages.

Four things to do this quarter

01

Open channel performance reporting

Look at where PMax actually spends. If most of it is Display and YouTube while you thought you were buying Shopping, you found your leak.

02

Exclude brand terms from PMax

Permanently. Then stop grading it on numbers it did not earn, and run brand through a Search campaign you can see.

03

Test AI Max against a holdout

If it was switched on for you, do not just accept it. Judge it on cost per acquisition, not revenue, because revenue is the number that goes up while you lose.

04

Keep Shopping funded

It is the cheapest efficient traffic in most ecommerce accounts, and the one most stores underfund.

Why brand traffic is the real split

A branded search is someone who already knows you. When PMax is allowed to serve that search, the sale lands in its return column and makes the campaign look like an acquisition engine. It was harvesting.

  • Brand cannibalization is argued here as a mechanism, not measured. No source has published a clean percentage, so none is claimed.
  • Every figure on this page is published data about companies I have no relationship with.
  • No client accounts are shown, in the video or here.

PMax vs Search FAQ

Is Performance Max better than Search for ecommerce?

Neither wins outright. On 2026 ecommerce benchmarks Search costs more per click and converts better, PMax is cheaper per click with a higher reported return, and Shopping is cheapest. They do different jobs, so the useful question is what each campaign is allowed to serve, starting with your brand terms.

Should I exclude brand terms from Performance Max?

In most ecommerce accounts, yes. Branded searchers already know you, so conversions PMax takes from them inflate its reported return without adding customers. Run brand through a Search campaign you can see and control instead.

Is AI Max worth turning on?

Treat it as a test, not a default. Across 250+ retail campaigns, Smarter Ecommerce found median revenue up 13% and median CPA up 16%, with outcomes from +42% to -35%. Run it against a holdout and judge it on cost per acquisition.

What is happening to Dynamic Search Ads?

Google’s ads liaison has confirmed Dynamic Search Ads are being deprecated and folded into AI Max. No timeline was given, so plan for the move rather than waiting on a date.

Sources

Captured September 13, 2026.

  1. 14% more conversions or conversion value at a similar CPA or ROAS, rising to 27% for campaigns still on exact and phrase match: Google’s AI Max claim, as reported by Search Engine Land, which noted the figure excludes retail.
  2. Median revenue +13%, median CPA +16%, outcomes from +42% to -35%, and the “essentially a coin toss” summary: Mike Ryan, Smarter Ecommerce, 250+ retail campaigns.
  3. Ecommerce benchmarks by campaign type (Search $1.42 CPC, 2.8%, 3.4x; PMax $0.82, 1.9%, 4.8x; Shopping $0.68, 1.4%, 5.1x): CorePPC 2026.
  4. PMax control additions (channel performance reporting, search term visibility, 10,000 campaign-level negatives, doubled search themes): Google product documentation and release notes.
  5. Dynamic Search Ads folding into AI Max: confirmed by Google’s ads liaison, no timeline given.

Related

Performance Max ManagementPMax campaigns run with asset-group discipline.Ecommerce PPC & Google Ads AgencySearch, Shopping and Performance Max for ecommerce brands.Free Google Ads Creative AuditScore your Google Ads against a competitor in minutes.

See your Google Ads next to a competitor’s

The free Google Ads creative audit pulls every ad you and a competitor are running from the Ads Transparency Center and emails you the comparison in about a minute. If you want me on your actual account, use the form below.

← Back

Thank you for your response. ✨