What Is Performance Max and What Still Depends on You Even Though AI Runs It

Performance Max is a Google Ads campaign type that serves ads across Search, Display, YouTube, Gmail and Maps from a single budget and a single conversion goal, letting machine learning decide where and to whom each combination is shown. By early 2026 it already generates close to 45% of all Google Ads conversions, according to DigitalApplied [1], but it still depends on the signals, the feed and the creatives you give it.

What Performance Max is and how it differs from a classic Search campaign

The advertiser supplies asset groups — up to 15 headlines, 5 descriptions, 20 images and 5 videos per group — and Google automatically assembles the optimal combination for each placement and user [1]. The more quality assets you provide, the more room the system has to build those combinations without falling back on generic assets.

What the AI decides inside a Performance Max campaign

What the AI does not decide on its own is the business goal: if the system optimises towards lead volume without you telling it which leads are worth more, it will end up chasing quantity, not profitability. A real case from a B2B SaaS account: after an aggressive expansion of Performance Max, leads doubled but profit fell 40% because the algorithm was optimising for volume, not value [4].

Which signals you still need to provide so the AI optimises well

Audience signals. They act as suggestions for the system, not hard restrictions [3]. Adding first-party customer lists as a signal can multiply conversion volume by 2 to 7 times compared with leaving the campaign with no signal at all, according to Pace Ads [5].

Up-to-date product feeds. In campaigns with a catalogue, a feed with outdated prices and stock confuses the system as much as it would confuse a shopper who sees an out-of-stock product in the ad.

Conversion data with business value, not just events. The case cited above was solved by importing CRM data with the real value of each closed deal, prioritising “opportunity won” over “lead created” as the optimisation signal [4]; ROAS recovered within 60 days, generating fewer but higher-value leads.

When it makes sense to use Performance Max and when it does not

One sign that Performance Max is not ready for your account: if you still do not have consistent conversion data from the last 30-60 days, the system has nothing to learn from. In that case, a well-structured classic Search campaign usually gives better visibility of what is working while you build up history.

How to interpret the channel report of a PMax campaign

Review the channel split every week, not every day: attribution data in Google Ads can take time to consolidate, and decisions taken on partial 24-48 hour data tend to overreact to noise, not to a real trend.

Common mistakes that make Performance Max perform worse than Search

• Not adding your own audience signals. Without them, the system starts from zero in every learning cycle, when it could start from your real customers [5].
• Optimising for lead volume instead of value. A real case showed leads doubling with 40% less profit by chasing this wrong metric [4].
• Outdated product feed. Incorrect price or stock confuses the system as much as the end user.
• Generic or insufficient assets. Fewer quality headlines and images reduce the system’s room to assemble winning combinations.
• Judging the campaign with less than 30 days of data. The system needs time and history to stabilise automated bidding.

How to cross-reference PMax data with the rest of your account from a single chat

In boostiq you can request a channel breakdown of a PMax campaign alongside the rest of your Google Ads account in the same conversation, without downloading or cross-referencing spreadsheets. If you want to catch up on this quarter’s Google Ads updates, the full guide is there. More context on campaign optimisation in the Central Hub.