Performance Max

What it is

Performance Max, often shortened to PMax, replaces manual placement and keyword control with a machine learning system that serves across Google surfaces. You build asset groups containing headlines, descriptions, images, logos and videos, optionally attach a product feed, and define listing groups, audience signals and search themes. The system then decides which combinations to show, where and to whom, optimising towards the conversion goals and values you set.

Why it matters

It concentrates a large share of Google spend into a campaign type where the advertiser controls inputs rather than placements, so input quality becomes the main lever. For discovery, it broadens reach into browsing and video surfaces that keyword campaigns do not cover, which can surface products to people who were not searching. It also changes measurement practice, because reporting is less granular than Search or Shopping campaigns and needs deliberate structure to interpret.

How it works

Practitioners feed the campaign strong conversion data, usually with values or offline conversion imports, so bidding optimises to profit rather than raw volume. Structure choices do the heavy lifting: splitting asset groups or campaigns by product category, margin or brand, using custom labels from the feed, adding brand exclusions and account level negatives, and supplying search themes and first party audience signals. Creative is refreshed based on asset performance labels, and results are read alongside brand and non brand Search to check incrementality rather than in isolation.

When it applies

It applies when an advertiser wants broad coverage across Google surfaces with automated bidding and has enough conversion volume for learning, particularly in ecommerce and lead generation. It is less suitable when strict placement control, tight keyword governance or very low conversion volumes make automation hard to steer.

Examples

• A retailer splits Performance Max into three campaigns by margin band using custom labels, then sets different target ROAS values for each.
• A lead generation business imports qualified lead and closed sale values from its CRM so bidding favours enquiries that convert offline.
• A brand applies a brand exclusion list and account level brand negatives to stop Performance Max absorbing cheap brand traffic and inflating reported ROAS.

How it is measured

• Conversions, conversion value and ROAS or cost per acquisition by campaign and asset group
• Incrementality of new customer acquisition, for example new customer share or lift tests against brand and non brand Search
• Asset performance ratings and coverage, plus share of spend going to Shopping versus other surfaces where reporting allows
• Search term and search category insights reviewed for waste, alongside brand versus non brand spend split