Google Search vs Performance Max: what are you actually giving up?

What each campaign type hands over

In a Search campaign you pick the questions: the keywords, the match types, the negatives. Google decides which auctions to enter and what to bid. In a Performance Max (PMax) campaign you pick a goal, upload assets and offer audience hints, and Google decides almost everything else. Google's own help page describes it as "a goal-based campaign type that allows you to access all Google Ads inventory", spanning YouTube, Display, Search, Discover, Gmail and Maps.

That is a real trade, not a trick. The useful question is how much of the old visibility you keep. On the Google help pages we read on 4 October 2026, the picture is better than it was at launch and still partial.

What you can see

• Channel breakdown. Google lists a channel performance breakdown among its reporting options, so you can see how results split across its inventory rather than one blended number.
• Search terms. There is a Performance Max search terms report showing "the search terms that triggered your Performance Max ads", with data available from March 2023. It does not include store visits or store sales conversions.
• Asset groups and assets. Reporting shows top-performing asset combinations (the six best in each category) and asset-level metrics. Google warns that conversions are not split across the parts of an ad, so asset-level conversions will not add up to the campaign total.

What you can steer

• Brand exclusions. They stop a campaign serving on branded queries you want to avoid, but "apply to only Search, Shopping, and YouTube search inventory". In PMax you can still let Shopping ads show for excluded brands if you choose.
• Negative keywords. Account-level negatives apply to all Search and Shopping inventory, including PMax. Google states they are "applicable to Search and Shopping inventory only", so Display and YouTube placements need content exclusions instead.
• Search themes. You can hint at the queries you want, but they add to what Google predicts; they do not fence it in.
• Final URL expansion. Google can send people to other pages on your site and generate dynamic assets. You can opt out through URL exclusions, though the campaign's own final URL cannot be excluded.
• Conversion goals. You can bid to account-default goals or pick goals for a single campaign. Whatever you mark as the goal is what the machine will chase, including the junk if your tracking counts junk.

Search is automating too. Google recommends pairing broad match with Smart Bidding, and its brand exclusions page notes that Search brand settings are moving into an AI Max panel. The gap is narrowing from both ends, which is why the test below asks about your situation rather than the label on the campaign.

Search against Performance Max, side by side

The see, steer, learn test

Automation does not make marketing worse. It moves the places where you can intervene. Before choosing PMax, run three tests. Each needs a yes you can prove, not a yes you hope for.

1. See. Can you open a report and say which queries, channels and pages produced last month's enquiries or sales, and tie them to money in your bank account? If the best answer is a reported conversion count, you cannot see.
2. Steer. Can you stop spend on something you do not want? Your own brand name, a competitor, a suburb you do not serve, a page that should never receive paid traffic. If steering means asking Google to adjust, you cannot steer.
3. Learn. Will the campaign teach you something you can use outside it, such as the phrase that wins, the offer that converts or the suburb that responds? This needs enough conversions to form a pattern. Without them the system learns and you do not.

Three yeses and PMax is a reasonable bet. One no, and Search will usually serve you better until you fix it. Two or three noes: the campaign type is not your problem, your tracking is. Start with the tracking check.

Where AI-assisted checking helps: an agent can read a PMax search terms export of several thousand rows in minutes and sort it into brand, competitor, irrelevant and useful. A person still decides what "irrelevant" means for a physio on Illawarra Road, and what to exclude.

The brand search sum

The most common way PMax flatters itself is by harvesting people who were already looking for you. Here is an illustrative sum for a Newtown homewares shop with a product feed. These figures are invented to show the arithmetic.

• PMax spend: $3,000 in a month. Google reports $18,000 in conversion value. Reported return on ad spend: 6.0.
• A search terms review shows 40% of that value came from searches containing the shop's own name. That is $7,200 from brand and $10,800 from everything else.
• Return on spend from non-brand queries alone: $10,800 divided by $3,000 is 3.6.
• At a 40% gross margin (also illustrative), the reported figures suggest $7,200 of margin less $3,000 spend, so $4,200 of profit. Strip out brand and it is $4,320 less $3,000, so $1,320. Profit looks about two thirds lower.

That does not mean the brand traffic was worthless. Some people search your name because a YouTube or Display ad introduced you, and PMax deserves some credit for that. The truth sits between the two figures, which is why the sensible move is to set brand exclusions, run brand in its own Search campaign, and compare a few weeks of each. Google also states that a Search campaign with a keyword that exactly matches, or is spell-corrected to, a person's query is preferred over PMax, which gives a brand Search campaign a structural advantage.

Our broader point about reported returns is in Good ROAS, bad growth.

Which campaign type for which business

Read the table as a starting point for the first six months, not a verdict. Every example is an illustrative business type, not a named client.

If you prefer a tree, take these in order and stop at the first answer that applies.

1. Do you sell products from a feed online or in store? If no, go to 2. If yes, go to 4.
2. Does the customer search for your service by name or need (a dentist, a plumber, a physio)? If yes, use Search. If no, go to 3.
3. Is your offer visual or impulse-led (a bar, a boutique, a venue)? Use Meta and your Business Profile first, and consider TikTok as a test.
4. Can you pass see, steer and learn, and meet the conditions in the next section? If yes, PMax with brand exclusions. If not, Shopping or Search until you can.

Minimum conditions for Performance Max

If you cannot tick most of these, run Search and come back.

• Tracking you have tested. The conversions marked as primary are real outcomes (a booked appointment, a paid order), not page views or button clicks, and you have confirmed they fire. Primary conversions drive bidding, so a bad one trains the system on the wrong thing.
• Enough volume to learn from. Google's Performance Max pages we read state no minimum number of conversions. As a rule of thumb that is ours, not Google's, if you cannot expect around 30 real conversions a month, the system is deciding on very little evidence. At $1,500 a month and $50 an enquiry you would get 30 conversions in total, and split across three asset groups that is about 10 each.
• Brand exclusions and account-level negatives set before launch. Not after the first invoice. Run a separate brand Search campaign.
• Assets you can produce. Google's optimisation tips recommend at least 20 text assets (15 headlines and 5 descriptions), 7 images and 1 video per asset group. Where you supply less, Google can generate dynamic assets, and you may not like them.
• Patience and budget for six weeks. Google advises running new campaigns for at least six weeks and avoiding frequent budget, bid strategy or status changes during that period. If you will panic at week two, do not start.
• A margin you know. Reported return is not profit. You need your gross margin to read any PMax result.

When Performance Max is the right call

PMax is not a villain, and for some businesses Search alone leaves money on the table.

• You sell a catalogue. A homewares or furniture shop with a clean product feed gets Shopping placements that text ads cannot match.
• Search demand is thin. If few people type your product's name, Search cannot find them. PMax can introduce the product on YouTube, Display and Discover.
• You have several locations. Store goals can optimise for store visits, store sales, call clicks or direction clicks. They need your Business Profile linked, or chosen affiliate locations, and for some goals the profile must be eligible for store visits. The page we read does not state the eligibility threshold, so ask before you plan around it.
• Your Search campaigns have stopped finding new customers. Once exact and phrase terms are saturated and results flatten, a well-fenced PMax campaign is a legitimate way to widen the net.

In every case the same conditions apply: tested tracking, exclusions, enough volume, six weeks.

What we would do first this week

• Open Conversions and confirm which actions are marked primary. Remove or demote anything that is not a real enquiry or sale.
• If you already run PMax, export the search terms and count how much spend and value comes from your own name. Do the sum from the brand search section with your numbers.
• Add brand exclusions and account-level negative keywords, or confirm they exist.
• Check that a brand Search campaign exists with exact-match keywords for your name.
• Run the see, steer, learn test and write down the answers. If you have one no, fix that before changing campaign type.
• Set a review date at least six weeks out before touching budgets or bids.

Where this advice is wrong

This is a reading of Google's help pages and general account behaviour, not a controlled experiment on your account. We do not know how any single PMax campaign will perform, and the right answer for a small business with a feed and good tracking can be PMax from day one.

Google changes names, menus and defaults often. For example, its brand exclusions page notes that brand settings move into the AI Max panel for Search. Check each setting in your own account. The conversion threshold is our rule of thumb, and the margin and spend figures in our worked examples are illustrative. Whether brand traffic was incremental can only be answered by a holdout test, and a small account may be too small to run one cleanly.

Finally, we audit read-only. We can tell you whether the controls above are in place, but we cannot see what Google does inside the auction.