Google Performance Max: A Practitioner's Honest Guide

Google Performance Max consolidates Search, Shopping, Display, YouTube, Gmail, and Maps into a single campaign type. The pitch is compelling: one campaign, AI-powered bidding, full Google inventory. The reality is more complicated. PMax gives Google enormous control over where your budget goes, and without the right asset configuration and audience signals, it will often spend on brand terms and branded searches you were already capturing cheaply, inflating apparent ROAS while doing little incremental work.

What PMax controls vs what you control

You supply asset groups: headlines, descriptions, images, logos, and videos. Google decides which combinations to show, on which placements, to which users, at what bids. You cannot select specific placements, exclude irrelevant Display websites at the campaign level, or see individual placement-level performance data. What you can control: audience signals (not targeting; these are hints), brand exclusions (critical: add them on day one), negative keywords at the account level, and asset group segmentation by product category or audience intent.

• Add brand exclusions immediately. Without them, PMax will cannibalize your branded search campaigns and report inflated ROAS
• Segment asset groups by product category, not by channel or audience; Google handles the channel allocation
• Provide strong audience signals: customer match lists, website visitors, and in-market segments relevant to your product
• Upload all asset formats. Campaigns without video assets default to auto-generated video which typically underperforms
• Set a target ROAS or target CPA that reflects what you actually need. Without a target, budget will concentrate on the easiest conversions (usually brand)

Where PMax wastes budget and how to spot it

The most common budget leak is brand term spend. If you don't have brand exclusions and a separate brand campaign, PMax will bid aggressively on your brand name, generate high-ROAS conversions that would have happened anyway, and make the campaign look like it's performing well. Run a search terms report; filter by brand terms and look at what percentage of conversions came from users who were already looking for you by name. If it's above 40%, you have a brand bleed problem. The second leak is low-quality Display placements. Check asset group insights for placement URLs if they're surfaced, and add account-level placement exclusions for mobile app categories and irrelevant content networks.

How to structure PMax alongside your existing campaigns

Don't replace all campaigns with PMax. The right structure for most advertisers is: a separate brand campaign with exact/phrase match, a standard Shopping campaign for your top SKUs where you want granular bid control, and PMax for prospecting and non-brand inventory. Set the standard Shopping campaign to higher priority than PMax so Google pulls product ads from the controlled campaign first. PMax then handles everything else. This structure lets you keep efficiency on core traffic while using PMax's cross-channel reach for net-new audience discovery. Run both for at least four weeks before comparing performance. PMax's learning period is longer than standard campaigns and early data is noisy.