Google Search vs Performance Max

Keyword-driven Search campaigns against Google's all-inventory automated campaign type, compared on controls, reporting, and cost drivers.

At a glance

Cost drivers

Neither campaign type costs anything to use; you pay per click or impression through the same Google Ads auction. What separates them is what actually drives your effective cost. In Search, you set the intent floor: your keywords decide which auctions you enter, so cost tracks query competition directly and a junk click usually traces back to a match-type or negative-keyword decision you can fix. In Performance Max, the intent mix is variable by design: it blends expensive high-intent search clicks with cheap awareness inventory, which can make blended CPC and cost per conversion look better than a like-for-like search comparison would.

The real levers in PMax are the Smart Bidding target, since your Target CPA or Target ROAS setting is the throttle everything else obeys, plus feed and asset quality, conversion data volume, and negative keyword hygiene. Brand traffic deserves special attention: unless you add brand exclusions, PMax will absorb brand queries that would have converted anyway, which flatters its reported numbers. Serving priority also shapes cost. Google's documentation says that when a query is identical to an eligible keyword in your Search campaign, the Search campaign serves; everything else is decided by Ad Rank, so a weak Search campaign quietly cedes traffic to PMax.

When to pick Google Search

Pick Search when you sell services, run B2B, or generate leads. These businesses live and die on query intent, and Search is the only campaign type that lets you name the exact queries you are willing to pay for. The full search terms report means every wasted dollar has a visible cause, and negatives work across all of it. If your conversion volume is thin, which is true for most local and B2B advertisers, Search also degrades more gracefully, since manual and enhanced bidding still function where Smart Bidding would be guessing.

Search is also where I keep regulated and reputation-sensitive categories. You control the ad text, you control the queries, and nothing gets auto-assembled into a context you did not choose. For a small account, one tight Search campaign with exact and phrase keywords remains the highest-control way to spend the first advertising dollars.

When to pick Performance Max

Pick PMax when you run e-commerce with a product feed. Shopping inventory is where it earns its keep, and the automation genuinely does find converting pockets of YouTube, Display, and Discover that you would never buy deliberately. The prerequisite is data: PMax needs steady conversion volume and accurate conversion values to optimize toward, so it belongs in mature accounts, not brand-new ones.

Treat your inputs as the new ad copy. Asset group quality, feed completeness, and audience signals are the only creative influence you have, so weak inputs produce weak automated output. And use the controls that now exist: brand exclusions on day one, negative keywords for junk queries, and the channel performance report to see where the money actually went.

What most comparisons miss

Most PMax criticism online is dated. The black-box complaints were fair in 2023, but since January 2025 advertisers can add campaign-level negative keywords, with the limit raised to 10,000, and Google has shipped a full search terms report for PMax plus a channel performance report that breaks results out by Search, Shopping, YouTube, Display, Discover, Gmail, and Maps. The caveat still worth knowing: PMax negative keywords apply only to Search and Shopping inventory, so placement exclusions remain the tool for junk on the other channels.

The second miss is that PMax failure is usually a data problem, not a product problem. Smart Bidding amplifies whatever conversion signal you feed it. Accounts with loose conversion tracking, or lead-gen accounts counting form fills that never become revenue, teach the machine to buy the wrong thing at scale.

Third, PMax grades its own homework if you let it. Without brand exclusions it takes credit for brand demand, and without the channel report you cannot see the mix shift. I do not credit a PMax campaign until brand is excluded and at least a month of channel-level data is in.