Performance Max runs your ads across every Google property — Search, Shopping, YouTube, Display, Gmail, Discover, and Maps — from a single campaign. Google’s AI handles bidding, placement, and audience targeting automatically. Your job is feeding it the right inputs: conversion goals, creative assets, product data, audience signals, and budget.
That’s the pitch. The reality is more nuanced. PMax works well when you give the algorithm enough data, clear goals, and strategic guardrails. It falls apart when advertisers treat it as a set-and-forget solution. And in 2026, the campaign type has evolved significantly from the black box it was at launch — with campaign-level negative keywords, search term visibility, channel performance reporting, and new customer acquisition goals now available.
This guide covers how to set up, structure, optimize, and scale Performance Max campaigns based on how the system actually works in 2026, not how it worked in 2022.
What Performance Max Does (and What Changed in 2025-2026)
Performance Max replaced Smart Shopping and Local campaigns in 2022. From day one, its design philosophy was simple: give Google’s AI a budget, a goal, and creative assets, and let machine learning figure out where to show your ads and how much to bid.
The system works by analyzing real-time auction signals — user device, location, time of day, search history, browsing behavior — and deciding, for each individual auction, whether to bid and how much. It does this across all Google networks simultaneously, which is something no manual campaign management can replicate at scale.
But the early versions had serious control gaps. No negative keywords. No visibility into which channels drove conversions. No way to tell if PMax was cannibalizing your branded search traffic. Advertisers were essentially handing Google their budget and hoping for the best.
The 2025-2026 updates changed this meaningfully. Here’s what you can now control that you couldn’t before:
Campaign-level negative keywords — You can now add up to 10,000 negative keywords per PMax campaign, applying to both Search and Shopping inventory. This was the single most requested feature since PMax launched. You can filter out low-intent queries (“free,” “DIY,” “cheap”) and prevent brand cannibalization without needing to go through a Google rep.
Search themes — Expanded from 25 to 50 per asset group. Search themes are PMax’s version of keyword guidance. They tell the algorithm which search query categories are relevant to your business. Google now shows a “usefulness” indicator that reveals whether your search themes generate extra traffic beyond what the algorithm would have found on its own.
Channel performance reporting — You can now see which Google networks (Search, Shopping, YouTube, Display, etc.) drive your conversions. This makes budget allocation decisions data-driven instead of speculative.
New Customer Acquisition goal — You can set higher conversion values for first-time customers versus returning buyers. This directs the algorithm to prioritize growth rather than just optimizing for the easiest conversions (which are often remarketing to existing customers).
Ad Rank priority model — Before late 2024, PMax automatically won auctions over Standard Shopping when targeting the same products. Google replaced this with an Ad Rank model where the better ad — based on bid, quality, and expected asset impact — wins, regardless of campaign type. This change made running PMax alongside Standard Shopping a viable strategy for the first time.
Setting Up a Performance Max Campaign: The Decisions That Matter
Most setup guides walk through every screen in the Google Ads interface. That’s useful, but it buries the decisions that actually affect performance under procedural steps. Here are the choices that make or break your campaign.
Conversion Goals: Fewer Is Better
PMax optimizes toward whatever conversion actions you tell it to. If you have three primary conversion goals — purchases, form submissions, and newsletter signups — the algorithm will chase whichever ones are cheapest to acquire. That usually means newsletter signups dominate, not purchases.
In 2026, Google recommends streamlined conversion setups: fewer, higher-value goals lead to better optimization. For ecommerce, your primary conversion should be purchases with revenue values attached. For lead generation, it should be qualified form submissions or phone calls — not page views, not PDF downloads.
Remove conversion actions that don’t represent actual business value before launching your campaign. You can keep them as observation-only secondary conversions, but don’t let the algorithm optimize toward them.
Budget: The Minimum That Actually Works
Google suggests a daily budget of at least three times your target cost per acquisition. If your target CPA is $50, start with at least $150/day. Campaigns with insufficient conversion volume — fewer than 30 conversions per month — struggle to exit the learning phase and deliver inconsistent results.
PMax runs across multiple channels simultaneously. A $30/day budget spread across Search, Shopping, YouTube, Display, Gmail, and Discover doesn’t give the algorithm enough data in any single channel to learn effectively. Underfunding is the most common reason PMax campaigns underperform.
If your budget is limited, concentrate the campaign on fewer goals and consider narrower geographic targeting to accumulate conversions faster in a smaller pool.
Bidding Strategy: Start Broad, Then Constrain
PMax offers two bidding strategies:
Maximize Conversions — Drives the highest volume of conversions within your budget. Best for lead generation campaigns or accounts that need to build conversion data quickly.
Maximize Conversion Value — Drives the highest total conversion value (revenue) within your budget. Best for ecommerce accounts where not all conversions are worth the same amount.
Both strategies support optional targets. Maximize Conversions can include a target CPA; Maximize Conversion Value can include a target ROAS.
The proven approach: launch without a target constraint first. Let the algorithm spend freely for 2-4 weeks and accumulate 30-50 conversions. Then set a target based on your actual achieved performance — not your aspirational goal. Adjust targets gradually, no more than 15-20% every two weeks. Aggressive target changes reset the learning phase and cause performance swings.
Campaign Structure: The Hybrid Strategy That’s Working in 2026
The biggest structural decision for ecommerce advertisers is whether to run PMax alone or alongside Standard Shopping. In 2026, the hybrid approach — PMax plus Standard Shopping — is producing the most consistent results across serious ecommerce accounts.
Why Hybrid Works Now
The Ad Rank priority change made this viable. Before late 2024, PMax automatically outbid Standard Shopping for the same products, making it pointless to run both. Now, the stronger ad wins based on quality and bid, regardless of campaign type. This means Standard Shopping can compete fairly for specific product segments where you want tighter control.
The hybrid logic is straightforward. Standard Shopping acts as a scalpel: granular control over bids, full search term visibility, Shopping-only placements. PMax acts as a discovery engine: broad reach across all Google networks, automated audience finding, cross-channel optimization. Together, they cover both precision and scale.
A Practical Hybrid Structure
Standard Shopping campaign — Your highest-performing, highest-margin products. Manual bidding or Target ROAS. Full control over search terms and bids. This campaign protects your core revenue drivers with predictable performance.
Branded Search campaign — A separate Search campaign targeting your brand keywords. This prevents PMax from cannibalizing branded traffic and inflating its reported ROAS. Apply brand exclusions in PMax to keep it out of branded auctions.
PMax campaign (hero products) — Your proven bestsellers with strong conversion data. Set a specific ROAS target. Full asset groups with headlines, descriptions, images, and video. Audience signals using customer lists and high-intent custom segments.
PMax campaign (catalog/discovery) — The rest of your product catalog. Moderate ROAS target. This campaign’s job is finding new products that could become heroes and reaching new audiences across all Google networks.
Accounts with lower conversion volume should simplify: one Standard Shopping campaign for core products and one PMax campaign for everything else. The goal is enough conversions per campaign for the algorithm to optimize effectively — splitting too thin starves both campaigns of data.
The Feed-Only Question
“Feed-only” PMax — uploading your product feed but skipping text, image, and video assets — used to be a workaround for advertisers who wanted PMax to behave like a Shopping-only campaign. In 2026, this workaround is increasingly unreliable. Multiple advertisers report that feed-only PMax campaigns now leak budget into YouTube, Display, and Search placements even without uploaded assets.
If you need guaranteed Shopping-only ad spend, use Standard Shopping. If you run PMax, commit to full asset groups. Internal testing shows manually created videos outperform auto-generated ones by 25-40%, and brands that provide at least one real video see up to 12% more total conversions compared to text-and-image-only asset groups.
Product Segmentation: The Hero/Villain Framework
Not every product in your catalog deserves the same ad spend. The Hero/Villain/Sidekick/Zombie framework, popularized by ProductHero’s Labelizer tool, gives ecommerce advertisers a structured way to allocate budget based on actual product performance.
Heroes — Your best performers. High ROAS, strong conversion rates, consistent volume. These products get the majority of your budget and the most aggressive ROAS targets.
Sidekicks — Products showing potential but not yet hitting targets. Maybe they have a high ROAS but low click volume, or good conversion rates on limited traffic. These get a moderate budget allocation. The goal is giving them enough exposure to either graduate to Hero status or reveal themselves as Villains.
Villains — Products that consistently spend budget without converting profitably. These get the tightest budget constraints or are excluded from paid campaigns entirely.
Zombies — Products that get almost no impressions or clicks at all. The algorithm ignores them because they lack data. These need dedicated attention — often in a separate campaign with broader targeting — to get initial traction.
Implement this framework using custom labels in your Google Merchant Center feed (custom_label_0 through custom_label_4). Assign each product a performance label, then create separate PMax campaigns or asset groups with different ROAS targets for each segment.
Some advertisers are moving beyond this static framework. Dynamic segmentation tools now shift products between categories automatically based on real-time signals — stock levels, competitor pricing, margin changes, conversion trends. The principle is the same (treat different products differently), but the execution adapts continuously.
Asset Groups: What Actually Moves Performance
Asset groups are collections of creative elements that PMax combines into ads across all placements. Each asset group should revolve around a specific product category, audience segment, or business objective — not your internal organizational structure.
Fill Every Slot
PMax has creative slots for a reason. The algorithm tests different combinations of headlines, descriptions, images, and videos across placements. The more variety you provide, the more combinations it can test, and the faster it finds winners.
Provide at least 11 of the 15 available headlines (30-character limit each). Mix keyword-focused headlines with feature highlights, offers, and social proof. Make at least one headline very short (under 15 characters) for mobile placements where space is tight.
Write 4-5 descriptions (90-character limit). Each description should highlight a different angle: price advantage, product quality, shipping speed, customer reviews, brand credibility.
Upload images in all three required aspect ratios (1.91:1 landscape, 1:1 square, 4:5 portrait). Use high-quality product photography and lifestyle images. Avoid text overlays — they reduce ad approval rates and compete with Google’s own headline overlays.
Upload at least one real video per asset group. If you don’t upload video, Google auto-generates one from your static assets. These auto-generated videos consistently underperform custom content.
Audience Signals: Guidance, Not Targeting
Audience signals tell PMax who your ideal customers look like. They’re suggestions to the algorithm, not hard targeting constraints. Google will show your ads to people outside your specified signals if it predicts they’re likely to convert.
Start with your strongest first-party data: customer email lists (especially high-value repeat buyers), website visitors (especially cart abandoners and converters), and CRM data if you can integrate it. Layer in custom segments built from your top 25-50 converting search terms.
For competitor conquesting, add custom segments based on competitor brand names and URLs. This works well when competitors serve a similar audience at a similar price point.
Create separate asset groups for different audience signals when you want to test which signals drive the best results. PMax’s standard reporting doesn’t break down performance by individual signal — isolating signals into separate asset groups is the workaround.
Search Themes: PMax’s Keyword Layer
Search themes tell the algorithm which search query categories are relevant to your asset group. You now have 50 slots per asset group. Don’t waste them on broad, vague terms like “clothing” or “marketing software.” Use specific, high-intent phrases that mirror your top-performing search terms from other campaigns.
Google’s “usefulness” indicator shows whether each search theme generates incremental traffic. If a theme shows low usefulness, the algorithm was already finding those queries without the hint. Replace low-usefulness themes with more specific alternatives.
Optimizing the Product Feed
For ecommerce PMax, the product feed is your most important optimization lever. PMax pulls product titles, descriptions, images, prices, and availability directly from Google Merchant Center. Weak feed data limits what the algorithm can do, regardless of campaign structure or bid strategy.
Titles
Product titles are the single most influential feed attribute for search relevance. Structure them as: Brand + Product Name + Product Type + Key Attributes (color, size, material). Front-load the most important details because some ad formats truncate titles.
A title like “Nike Air Max 270 Men’s Running Shoe – Black/White, Size 10” gives the algorithm far more to work with than “Men’s Shoes.”
Images
Use multiple high-resolution images (minimum 1000×1000 pixels). Show the product from multiple angles on clean backgrounds. No watermarks, no promotional text, no collages — these violate Google’s image policies and lead to disapprovals. Google Merchant Center rejects approximately 7% of products due to data errors, and image violations are among the most common.
Custom Labels for Strategic Control
Custom labels (custom_label_0 through custom_label_4) don’t affect ad display but give you powerful internal segmentation capabilities:
- custom_label_0: Margin tier (HighMargin, MidMargin, LowMargin)
- custom_label_1: Performance bucket (Hero, Sidekick, Villain, Zombie)
- custom_label_2: Seasonality (Spring, Summer, Evergreen)
- custom_label_3: Inventory status (InStock, LowStock, Clearance)
- custom_label_4: Business priority (NewLaunch, CoreRange, Discontinued)
These labels let you create product groups within PMax campaigns, set different ROAS targets by segment, and exclude products strategically. Each label can hold only one value per product, and new or edited labels take 24-48 hours to propagate to Google Ads.
The Incrementality Problem
PMax’s biggest blind spot isn’t its automation — it’s attribution. The campaign will happily bid on your branded search queries, convert people who were already coming to your site, and report that result as PMax-driven revenue. Your ROAS looks amazing. Your actual incremental revenue might be flat or declining.
This is why brand exclusions matter. Exclude your own brand name in PMax settings so the campaign stops bidding on branded queries that your branded Search campaign should handle. Compare PMax performance before and after applying brand exclusions. If ROAS drops significantly, a large portion of your “PMax revenue” was branded traffic you would have captured anyway.
For more rigorous incrementality testing: split your product catalog into two comparable segments (similar margin mix, similar demand). Run one segment in PMax and the other in Standard Shopping. Don’t overlap products, or you’re just measuring which campaign bid more aggressively, not which campaign type performs better.
Google also offers conversion lift studies for larger accounts — controlled experiments that measure the true incremental impact of PMax by comparing exposed and holdout groups. If your account is large enough to qualify, these studies provide the cleanest incrementality data available.
Performance Max for Lead Generation
PMax isn’t just for ecommerce. Lead generation accounts can use it effectively, but the setup requires different thinking.
The Junk Lead Problem
PMax optimizes for whatever conversion action you tell it to. If your conversion goal is “form submission” and 60% of your form fills are spam or unqualified, the algorithm is learning to find more spam. It’s doing exactly what you asked — just not what you wanted.
The fix: connect your CRM to Google Ads through enhanced conversions for leads or offline conversion imports. Feed back which leads actually became opportunities or customers. This gives the algorithm a signal for lead quality, not just lead volume. Accounts that implement CRM-to-Ads feedback loops consistently see lower cost per qualified lead, even if their total lead volume drops initially.
Budget and Conversion Volume
Lead generation campaigns typically have higher CPAs than ecommerce purchases, which means you need more budget to reach the 30-50 conversion threshold for effective optimization. If your target CPA is $100, you need at least $300/day to give the algorithm enough data.
If budget is tight, start with Maximize Conversions (no target constraint) to accumulate conversion data. After reaching 50 conversions, add a target CPA based on actual results.
Audience Signals for Lead Gen
Your strongest signals are high-quality leads from your CRM (not all leads — specifically the ones that converted to customers or opportunities). Upload these as Customer Match lists. Build custom segments from the search terms that historically produce qualified leads, not just any leads.
In-market audiences and detailed demographics help for top-of-funnel awareness, but first-party data from qualified leads drives the most efficient performance.
Performance Max for Local and Store Visits
PMax for store goals promotes physical business locations across Google Maps, Search, YouTube, and Display. Ads appear as promoted pins on Google Maps and Waze, showing business information — reviews, hours, photos — along with action buttons like “Call Now” and “Get Directions.”
Radius targeting adjusts dynamically based on user proximity to your store. Users closer to your location see your ads more frequently and with higher bid multipliers.
For local campaigns, make sure your Google Business Profile is verified, complete, and consistent with your website’s contact information. Link your Google Business Profile to your Google Ads account before launching. Location assets (formerly location extensions) are required for store visit tracking.
Tracking and Improving Performance
The Insights Tab
The Insights tab in Google Ads is your primary diagnostic dashboard for PMax. It shows search category breakdowns, audience insights, trending search queries, and asset performance. Check it weekly — it surfaces optimization opportunities that raw metrics don’t reveal.
Asset Performance Ratings
PMax rates each creative asset as “Low,” “Good,” or “Best” based on relative performance within its asset type. Review these ratings after at least 14 days of data collection. Replace “Low”-rated assets, but don’t delete them without having replacements ready — removing assets without adding new ones reduces the algorithm’s creative options.
The new “Ads Using Video” reporting segment (available in 2026) lets you compare conversions from ads that included video versus those that didn’t. Use this to quantify whether your video investment is paying off.
Search Term Insights
Search term reports for PMax now show queries grouped by theme and intent, with performance metrics for each group. You can identify which query categories drive conversions, which waste budget, and which represent new opportunities. The report also distinguishes between queries that matched your search themes and those the AI selected independently.
High-converting query themes should inform your search theme settings and headline copy. Low-performing or irrelevant query themes should be added as negative keywords — you can now do this directly from the report with a single click.
Frequently Asked Questions
Should I replace my Search campaigns with Performance Max?
No. Google designed PMax to complement Search campaigns, not replace them. Search campaigns with exact match keywords take priority (via Ad Rank) when a query matches your keyword. Run PMax alongside Search to capture broader queries and cross-channel opportunities that Search campaigns miss. Most successful accounts maintain separate branded Search, non-branded Search, and PMax campaigns.
How long does Performance Max take to optimize?
Plan for at least 4-6 weeks before evaluating performance. The algorithm needs time to test creative combinations, identify responsive audiences, and learn from conversion data. Making significant changes (budget, bidding targets, asset groups) during the learning phase resets the process. After the initial learning period, optimize incrementally — adjustments of no more than 15-20% every two weeks.
Is feed-only PMax still a viable strategy in 2026?
Increasingly not. Feed-only PMax campaigns — where you upload a product feed but skip text, image, and video assets — were designed to force Shopping-only placements. In 2026, multiple advertisers report that feed-only campaigns now leak budget into YouTube, Display, and Search placements even without uploaded assets. If you need guaranteed Shopping-only ad spend, use Standard Shopping instead.
How do I prevent PMax from cannibalizing my branded search traffic?
Apply brand exclusions in your PMax campaign settings. Create a separate branded Search campaign to capture brand-intent queries. Compare your overall account performance before and after implementing brand exclusions — a significant drop in PMax ROAS after exclusion typically means the campaign was claiming credit for branded conversions it didn’t generate.
What’s the minimum budget for Performance Max?
Google recommends at least three times your target CPA per day. For ecommerce accounts targeting a $30 CPA, that’s $90/day minimum. For lead generation with a $100 CPA target, that’s $300/day. Campaigns with fewer than 30 conversions per month typically struggle to exit the learning phase and deliver inconsistent results. If budget is limited, consider narrower geographic targeting to concentrate conversions.
Can I use Performance Max without a product feed?
Yes. Service businesses, SaaS companies, and lead generation accounts run PMax without a Merchant Center feed. The campaign will deliver ads across Search, Display, YouTube, Gmail, and Discover using your text, image, and video assets. You lose Shopping placements without a feed, but all other networks remain active.
