Performance Max Campaigns: Run Them Without Guesswork

Performance Max campaigns promise a lot. One campaign runs across Search, YouTube, Display, Discover, Gmail, Maps and Shopping, and Google’s AI handles bidding, budget, audiences and creative. Then you look at the results. Spend climbs. Conversions look healthy, but many come from people who searched your brand name. Leads arrive, and sales says many of them are junk. You can’t easily see where the money went.

That is the cost of running Performance Max loosely. You pay for clicks on your own name. Bad leads teach the algorithm to bring more bad leads. The campaign can end up competing with your Search campaigns for the same queries. And changes made out of frustration can reset the learning period, which takes about six weeks to begin with.

Kyō, the AI marketing manager at Kyōsey AI, takes this on with her Google Ads expert. First they decide whether Performance Max fits your business at all. If it does, they plan it on the right conversion data, keep it away from your brand and Search traffic, give it assets that work, and watch it patiently. Everything they prepare comes to you to approve. Nothing goes live without your OK.

Why Performance Max Campaigns Go Wrong

Performance Max rarely fails because Google’s AI is weak. It fails because of what it is given and how it is managed. The common problems are predictable:

• It’s launched too early. The account doesn’t have enough conversion volume, or it counts the wrong conversions. The AI then optimises hard toward the wrong goal.
• It’s fed shallow signals. If a form fill counts the same as a qualified sales opportunity, the campaign learns to find form fillers.
• It eats the brand. Branded searches convert cheaply. PMax will take them if you let it, and its reported results look better than they really are.
• It’s touched constantly. Changing budget, bid or status during learning resets it. Impatient management keeps it permanently unstable.
• It’s split into too many campaigns. Google allows up to 100 Performance Max campaigns per account but recommends consolidating. More campaigns usually means thinner data in each one.

None of these show up as an error message. They show up as a campaign that “sort of works” and a budget that never grows.

Performance Max Vs Search Campaigns: Who Wins The Query?

This is the question that causes the most overlap. Google’s help pages suggest that when a query matches a keyword in your Search campaign, Search wins. Experienced strategists read that more narrowly. Search is preferred when the keyword’s text is identical to the query. A phrase or broad keyword that merely matches it can end up competing with PMax on Ad Rank.

Kyō works from that narrower reading. She plans your Search keywords and your Performance Max campaign so each has a clear job, and she recommends brand exclusions and PMax negative keywords, with the steps to add them, so branded traffic doesn’t drift into PMax by default. Performance Max is meant to complement Search, not replace it, and she sets it up that way.

What Kyō Does with Your Performance Max Campaigns

Kyō covers Performance Max end to end, from “should we run it?” to the troubleshooting nobody enjoys. She works alongside Kyōsey AI’s tracking and creative experts; you can see the whole team on the Features page.

She decides whether PMax is the right tool. PMax suits businesses with concrete conversion goals, such as online sales, leads or store visits, that will trade channel control for reach and automation. Kyō checks your conversion volume, your tracking and your goals first. Sometimes the honest answer is “build Search first” or “one consolidated campaign, not four.”

She makes sure the fuel is right. For lead generation, she wants the campaign optimising toward qualified leads coming back from your CRM, not raw form submissions. The tracking expert connects to sources like HubSpot, CallRail and GA4 to check that the campaign learns from outcomes that matter.

She structures it for learning. She keeps campaigns consolidated where that helps the data, and separates them only for a real business reason, such as different margins, products or goals. For retailers, she plans listing groups and reviews the Merchant Center feed for problems. She also plans page feeds and URL expansion and exclusions so traffic lands on pages that convert.

She builds assets that perform, not assets that fill boxes. Google’s guidance for Excellent Ad Strength is at least 15 headlines and 5 descriptions, at least 7 images (3 landscape, 3 square, 1 portrait) and at least one video. To reach Excellent and stop Google generating its own AI videos, you need video in all three orientations. Kyō takes a more practical line. She aims for Good rather than Excellent, because Ad Strength isn’t used in the auction. A few strong assets plus one real video beat filling every field. Supplying your own video also limits spend on auto-generated YouTube videos. She keeps logos and business information at the campaign level, where Google now expects them.

She respects the learning period. New campaigns need about six weeks to exit learning. After a significant change, she allows one to two weeks, or a full conversion cycle, before judging it. She plans changes so they don’t keep resetting learning, and she tells you why she’s holding off rather than tinkering.

She reads the reporting properly. She reviews asset, channel and search-term reports to show where the campaign serves, which queries it buys and how much of that is branded.

She troubleshoots. She works through the cause of each problem, not just the symptom, including:

• campaigns not serving, or low traffic
• sudden swings in spend or performance
• setup errors and Merchant Center issues
• disapprovals and image-policy problems

What You Get, Concretely

The value is simple: less money spent on clicks you’d have won anyway, less time lost to resets and guesswork, and a campaign that improves because it learns from the right signals.

The Judgment Behind The Automation

Google’s own documentation leans toward automation and toward Performance Max. Much of it is right. But an experienced strategist knows the trade-offs. PMax shows less about placements than Search does, the AI controls most of the targeting, and some widely repeated advice is contested. Audience signals are a good example. Kyō recommends the cheap, high-quality ones, such as your customer lists and site converters, and doesn’t sink hours into elaborate signal building that the evidence doesn’t clearly support.

That balance comes from how Kyōsey AI works. Kyō’s skills are the methods of experienced strategists, drawn from more than 100 person-years of agency experience and written down so she follows them every time. She also checks specs, limits and feature availability against Google’s current documentation before anything ships, because Performance Max changes often. Generative image tools, for instance, are gated by account history and vertical and are rolling out gradually.

Who This Is for

• Business owners who were told to “just turn on Performance Max” and want to know whether it’s paying off.
• In-house marketers who need PMax and Search to work together, not fight each other.
• Agencies that want consistent, careful PMax builds and monitoring across many accounts without adding headcount.
• Ecommerce brands selling through Merchant Center, and local or lead-gen businesses, such as a regional HVAC company or a multi-location clinic, that need qualified leads rather than volume.

How to Get Started

Kyō connects to your Google Ads account and the data around it: GA4, Search Console, Shopify, CallRail, HubSpot and more. She reviews your current setup and tells you what she’d do, step by step. Your data stays yours, and nothing goes live without your OK.

It’s elite agency performance for $6/day. The quickest way to see what she’d do with your Performance Max campaigns is to book a demo.