Performance Max Campaigns: Complete Setup & Strategy Guide

Performance Max is Google's AI-driven, all-in-one campaign type, and in 2026 it is the anchor of most serious Google Ads accounts. One campaign serves your ads across Search, Display, YouTube, Gmail, Discover, and Maps, with Google's algorithm deciding where your budget goes based on conversion probability. For years, advertisers called it a black box. That excuse is dead. PMax now has campaign-level negative keywords, customer list exclusions, brand controls, full placement reporting, and asset experiments, which means the advertisers who win are the ones steering it properly. This guide covers how Performance Max actually works, when you should and should not use it, the complete setup process, asset requirements, audience signals, bidding strategy, campaign structure, and the weekly optimization routine that keeps it performing. By the end, you should be able to build a PMax campaign the right way and avoid the mistakes that quietly waste most PMax budgets.

1What Is Performance Max?

Performance Max is a goal-based campaign type that runs your ads across every Google channel from a single campaign. Instead of building a Search campaign, a Display campaign, a YouTube campaign, and a Shopping campaign separately, you build one PMax campaign, load it with assets, and Google's AI decides which channels, placements, and combinations will drive the most conversions for your goal.

You give the system three things. Your conversion goals and values. Your creative assets (headlines, descriptions, images, logos, videos, and a product feed if you sell physical products). And your audience signals, which tell the algorithm where to start looking. From there, Smart Bidding handles every auction and the asset system assembles your ads on the fly for whatever placement opens up.

Performance Max launched in late 2021 as the replacement for Smart Shopping and Local campaigns. Since then it has absorbed more and more of the platform, and for most ecommerce businesses it is now the default campaign type. For lead generation, it has matured into a legitimate option as well, as long as your conversion tracking is airtight.

Question to Answer:

Do you understand that Performance Max is not a placement or a network, but a goal-driven campaign that spans every Google channel at once?

2How Performance Max Works

The core mechanic of Performance Max is simple to describe and hard to do well. You define the outcome, and Google's AI works backward from that outcome to find the people, placements, and creative combinations most likely to deliver it.

Here are the moving parts.

• Channels. One PMax campaign is eligible to serve across Google Search, the Google Display Network, YouTube, Gmail, Discover, and Google Maps. You do not pick the channels. The algorithm allocates budget across them based on where conversions are happening.
• Asset groups. The PMax equivalent of an ad group. Each asset group holds a themed collection of headlines, descriptions, images, logos, videos, and audience signals. Google mixes and matches the assets inside a group to build ads for every placement.
• Audience signals. Suggestions, not targeting. You hand the algorithm your best audiences (customer lists, website visitors, in-market segments) as a starting point, and it expands well beyond them to find new converters.
• Smart Bidding. PMax only supports Maximize Conversions (with optional Target CPA) and Maximize Conversion Value (with optional Target ROAS). Every bid in every auction is set by the algorithm using your conversion data.
• The feedback loop. This is the part most advertisers underestimate. PMax learns entirely from your conversion data. If your tracking is accurate, the system optimizes toward real business outcomes. If your tracking is broken or polluted with junk conversions, PMax amplifies the bad data at scale.

That last point is the single most important sentence in this guide. Performance Max lives and dies by the quality of its feedback loop. Before you spend a dollar on PMax, your conversion tracking needs to be right, your conversion values need to reflect reality, and your primary conversions need to be actions that actually matter to your business.

Question to Answer:

Is your conversion tracking accurate enough that you would trust an algorithm to spend your entire budget based on it?

3Performance Max in the 2026 Power Pack

Google's current framework for its AI campaign types is what it calls the Power Pack, introduced at Google Marketing Live in 2025. It has three pieces, and understanding how they divide the work helps you decide where PMax fits in your account.

• Demand Gen. Creates demand. Visually rich campaigns across YouTube, Shorts, Discover, Gmail, and now the Google Display Network. This is where awareness and social-style advertising lives, and it is where standalone Display campaigns are being migrated through 2027. I covered that transition in my Google Display Network guide.
• AI Max. Captures demand on Search. A suite of AI-powered Search features that expands your keywords with keywordless technology while keeping Search-level controls. It is Google's successor to legacy Search automation like Dynamic Search Ads.
• Performance Max. Orchestrates the full funnel. PMax sits across every channel simultaneously and pushes budget wherever conversion probability is highest at any given moment.

A few practical notes on how PMax relates to the campaign types you already run.

PMax and Search. PMax respects your existing exact-match Search keywords. If a query exactly matches a keyword in your Search campaign, the Search campaign generally wins the auction internally. That means a well-built Search campaign and a PMax campaign coexist fine, with PMax sweeping up the queries your Search campaigns do not cover. Keep your brand terms protected with brand exclusions in PMax so it does not cannibalize cheap brand traffic and claim credit for it.

PMax and Shopping. For ecommerce, PMax with a Merchant Center feed has replaced Smart Shopping entirely and is the default way to run product ads. Standard Shopping still exists, and it is the right choice when you want full manual control over products and bids, or when your conversion volume is too low for PMax to stabilize.

Question to Answer:

Do you know which job each Power Pack campaign is doing in your account, or are you running overlapping campaigns competing for the same conversions?

4When to Use Performance Max (and When Not To)

Performance Max is powerful, but it is not the right starting point for every account. Here is the honest breakdown.

Use Performance Max when:

• You run ecommerce with a Merchant Center feed. PMax is the default product advertising campaign in 2026. If you sell physical products online, you will almost certainly run PMax.
• You have solid conversion volume. Each PMax campaign needs roughly 30 or more conversions per month to stabilize and optimize. Below that, the algorithm is guessing.
• Your conversion tracking is accurate and value-based. Enhanced Conversions on, conversion values set, junk conversions excluded from your primary column.
• You have already maxed out Search. PMax is a strong scaling layer once your Search campaigns are profitable and you want incremental volume across other channels.
• You can feed it real creative. Multiple images, multiple videos, strong copy. PMax with two images and no video is running at a fraction of its potential.

Hold off on Performance Max when:

• You are brand new to Google Ads. Start with a Search campaign. Learn the platform, build conversion history, then layer PMax in. My Google Ads strategy guide covers the right build order.
• Your conversion tracking is not trustworthy yet. PMax amplifies whatever data you feed it. Fix tracking first.
• You generate low lead volume with long sales cycles. A B2B account with 8 conversions a month will struggle. Either feed PMax offline conversion data from your CRM so it can see which leads became revenue, or stay on Search until volume grows.
• Lead quality is already a problem. PMax optimizing toward form fills will happily find you more cheap, low-quality form fills. You need value-based conversion data before you point this machine at lead generation.

Question to Answer:

Does your account have the conversion volume and tracking quality PMax needs, or would a Search campaign serve you better right now?

5The 2026 Controls That Ended the Black Box Era

The biggest change in Performance Max over the past two years is not performance. It is control. Google has steadily handed back the levers advertisers spent years asking for, and if you are still running PMax like it is 2023, you are ignoring tools that meaningfully change results.

• Campaign-level negative keywords. You can now add negative keywords directly to PMax campaigns without filing a support request. Use them for brand safety, compliance risks, and clearly irrelevant queries. Do not use them to aggressively sculpt traffic the way you would in Search. PMax needs query freedom to find profitable demand your keyword research missed.
• Brand exclusions and brand guidelines. Exclude your own brand terms so PMax does not take credit for traffic that was coming anyway, and exclude competitor brands you do not want to appear against. Brand guidelines let you lock your colors, fonts, and logo treatment so AI-assembled formats stay on brand.
• Customer list exclusions. You can upload first-party customer lists and exclude them from a PMax campaign. This is how you force PMax to do pure new-customer acquisition instead of quietly remarketing to people who would have bought anyway. Pair it with your Customer Match lists.
• New customer acquisition and High Value Mode. The "Bid more for new customers" setting lets you tell PMax that a net-new customer is worth more than a repeat one, and value-based settings let you push bids harder toward users who look like your highest lifetime value customers.
• Full placement reporting. You can now see where your PMax ads actually showed, including Display and YouTube placements. Review it and apply account-level placement exclusions when you find budget going to low-quality apps and sites.
• Asset experiments. You can run structured creative tests inside PMax, comparing different creative angles (a discount-led concept versus a social-proof-led concept, for example) without tearing apart the campaign's learning.
• Search themes and reporting on query categories. Search themes let you suggest query territory to the algorithm, and the reporting now shows whether traffic came from your themes or from Google's AI expansion.

The takeaway is that "PMax is a black box" is no longer a reason to avoid it, and it is no longer an excuse for bad performance. The controls exist. The advertisers getting the best results are the ones actually using them.

Question to Answer:

Have you set up brand exclusions, negative keywords, and customer list exclusions on your PMax campaigns, or is the algorithm still spending with no guardrails?

6Step-by-Step: Setting Up Your First Performance Max Campaign

Here is the full setup process. This assumes your Google Ads account exists, your conversion tracking is running, and (for ecommerce) your Merchant Center feed is connected.

1. Confirm your data layer first. Enhanced Conversions turned on. GA4 linked. Conversion values assigned. Primary conversions limited to actions that matter (purchases, qualified leads). PMax will amplify whatever you feed it, so this step is not optional.
2. Create a new campaign and pick your objective. Sales for ecommerce, Leads for lead generation. Then choose Performance Max as the campaign type. If you have a Merchant Center feed, attach it here.
3. Choose one goal per campaign. Do not mix purchases and newsletter signups in the same campaign. One campaign, one business outcome. Mixed goals confuse the optimization and you lose the ability to read results.
4. Set your bid strategy. Ecommerce: Maximize Conversion Value, adding a Target ROAS once you have history. Lead generation: Maximize Conversions, adding a Target CPA once you have history. Set your initial target roughly 20% more conservative than your true goal so the algorithm has room to exit the learning phase, then tighten gradually.
5. Configure campaign settings. Locations and languages as usual. Turn off "Final URL expansion" if you need traffic restricted to specific landing pages, or leave it on and add URL exclusions for pages you never want ads pointing to (careers, support, blog categories that do not convert).
6. Enable new customer settings if relevant. If your goal is acquisition, turn on "Bid more for new customers" and upload your existing customer list both as the definition of an existing customer and, if you want pure acquisition, as an exclusion.
7. Build your first asset group around one theme. One product category, one service, one customer segment per asset group. Load every asset slot (full specs in the next section).
8. Attach audience signals to the asset group. Customer Match lists, website visitors, converters, and your best in-market segments. Signals are a starting point for the algorithm, not a targeting cage.
9. Add your brand exclusions and negative keywords. Exclude your brand terms (run them in a dedicated Search campaign instead) and add the obvious irrelevant negatives before launch, not after the budget is spent.
10. Set a realistic budget and launch. Give the campaign enough daily budget to realistically produce conversions every day. A $10/day budget against a $50 CPA target cannot learn. Expect a learning phase of one to two weeks, and resist the urge to touch anything during it.

Question to Answer:

Did you set up the guardrails (brand exclusions, URL rules, negatives, customer exclusions) before launch, or are you planning to add them after the budget is already wasted?

7Campaign Structure: The Over-Segmentation Mistake

The number one structural mistake advertisers make with Performance Max in 2026 is over-segmentation. The old Search mindset says more campaigns equals more control. With PMax, more campaigns means your conversion data gets split into smaller pools, and small data pools starve the algorithm.

The rule of thumb: every individual PMax campaign should be able to generate at least 30 conversions per month. If splitting your account into five PMax campaigns leaves each one with 8 conversions a month, you have built five campaigns that will never exit learning. Consolidate.

Here is how to think about structure.

• Split campaigns only when the business goal differs. Different ROAS targets, different conversion goals, or genuinely different economics (high-margin versus clearance products) justify separate campaigns. Different product categories alone do not. That is what asset groups are for.
• Use asset groups for themes inside one campaign. An apparel store runs one PMax campaign with asset groups for running gear, athleisure, and accessories. Each asset group gets its own creative and audience signals, but the conversion data pools at the campaign level where the bidding happens.
• Use listing groups to control which products serve where. For ecommerce, listing groups inside asset groups let you subdivide your feed without fragmenting your campaigns.

For advanced ecommerce accounts, the next level is what a lot of practitioners call campaign orchestration. Instead of fixed product groupings, you use custom labels in your Merchant Center feed to tag products dynamically based on live business data: margin, stock depth, price competitiveness, and historical ROAS. High-margin, in-stock, competitively priced products get labeled into a heavily funded PMax campaign. When stock drops or a competitor undercuts the price, the label updates and the product routes to a lower-budget campaign. Your spend follows your actual business conditions instead of a structure someone built six months ago. You do not need this on day one, but it is where serious ecommerce accounts end up.

Question to Answer:

Can every PMax campaign in your account realistically hit 30 conversions a month, or have you fragmented your data into campaigns that can never learn?

8Asset Group Requirements and Creative Specs

In a world where the media buying is automated, your creative is your targeting. The assets you provide determine which placements you are eligible for and who the algorithm decides to show your ads to. Filling every asset slot with strong, varied creative is the highest-leverage work you can do in PMax.

Here are the current specs and how to use each asset type.

Three creative rules that matter more than anything else.

1. Upload your own videos. If you do not provide video, Google auto-generates one from your images and text, and auto-generated videos consistently underperform real creative. You do not need a production crew. A clean 15-second vertical video shot on a phone, with your hook in the first 3 seconds, beats an auto-generated slideshow. Most viewers skip the moment they can, so the open has to carry the message. Google's built-in generative tools (including its Veo video model) can fill gaps, but authentic creative still wins.
2. Use human-centric imagery. Images showing real people using your product or service consistently outperform sterile product-only shots and generic stock photography. Lifestyle beats catalog.
3. Give the algorithm variety, not repetition. Fifteen headlines that all say the same thing in slightly different words give the system nothing to test. Mix offers, benefits, features, urgency, and social proof so PMax can match the right angle to the right user.

Asset Strength matters here too. Campaigns rated Excellent get access to more placements and tend to see better efficiency than ones limping along on minimum assets. Filling the slots is not busywork. It directly expands where you can serve.

Question to Answer:

Is every asset slot filled with varied, high-quality creative, including real video in all three aspect ratios?

9Audience Signals: Telling the Algorithm Where to Start

Audience signals are the most misunderstood part of Performance Max. They are not targeting. Adding an audience signal does not restrict your campaign to that audience. It tells the algorithm "people like this convert for me, start here," and then PMax expands far beyond the signal to find new converters who share the same patterns.

Because signals are a starting point, the quality of the seed matters enormously. Here is the priority order.

1. Customer Match lists of your best customers. Not your full email list. Your repeat buyers, your highest lifetime value clients, your closed-won deals. The algorithm models who to find based on who you give it, so give it your best, not your biggest.
2. Website converters and high-intent visitors. Purchasers, lead submitters, cart and pricing page visitors. Build these as remarketing audiences in GA4 and Google Ads. My Google Ads remarketing guide covers how to build every list type.
3. Custom segments built on high-intent search terms and competitor URLs. Tell Google the searches your buyers make and the competitors they consider.
4. In-market segments relevant to your offer. Useful as a supporting layer, weaker as a primary seed.

Skip broad affinity audiences as primary signals. "Sports fans" tells the algorithm almost nothing. A 2,000-person list of repeat customers tells it everything.

One more distinction worth knowing: search themes. Search themes are keyword-style suggestions you attach to an asset group to point PMax at query territory it might miss. Like audience signals, they are guidance rather than strict targeting, and they are especially useful for new products or services with no search history.

Question to Answer:

Are your audience signals built from your highest-value first-party data, or did you attach a couple of broad interest categories and call it done?

10Bidding Strategy for Performance Max

Performance Max supports exactly two bid strategies, each with an optional target. Picking the right one and ramping the target correctly is most of the game.

• Maximize Conversions (optional Target CPA). For lead generation. Start without a target so the algorithm can learn freely. Once you have steady volume and a stable cost per conversion, add a Target CPA close to your actual average and tighten it gradually.
• Maximize Conversion Value (optional Target ROAS). For ecommerce and any account with real conversion values. Same ramp: launch without a target or with a conservative one, then tighten as history builds.

The ramping rules that keep you out of trouble:

1. Start roughly 20% more conservative than your true goal. If your real target is a 400% ROAS, launch around 320 to 350. Give the system room to find volume, then walk the target up.
2. Change targets and budgets gradually. Moves bigger than about 15 to 20% at once can throw the campaign back into a learning period. Adjust, then wait 5 to 7 days before adjusting again.
3. Do not panic in the learning phase. The first one to two weeks will be volatile. Judge the campaign on rolling 30-day windows, not on yesterday.
4. Feed it value, not just volume. For lead generation especially, import offline conversion data from your CRM so the algorithm optimizes toward leads that became revenue instead of form fills that went nowhere. This single change is what separates PMax lead gen accounts that work from ones that drown in junk leads.

For the bigger picture on every bid strategy across all campaign types and how to transition between them, see my Google Ads bidding strategies guide.

Question to Answer:

Are you ramping your targets gradually from a conservative starting point, or did you demand your final ROAS on day one and choke the campaign?

11The Weekly Optimization Routine (and the Don't-Touch List)

Performance Max rewards patient, macro-level management and punishes constant tinkering. The optimization work is real, but it looks different from managing a Search campaign. Here is the weekly routine.

1. Review the Insights tab. Look at search categories, demand trends, and audience insights. This is where you find out what PMax has discovered about your buyers, and it often feeds your broader marketing strategy, not just this campaign.
2. Audit placement reporting. Check where your ads actually showed. Low-quality apps and clickbait sites get added to account-level placement exclusions. Same discipline as Display: the cheap clicks are only cheap if they convert.
3. Prune low-rated assets. In asset group reporting, replace anything rated Low with a fresh variant testing a genuinely different angle, not a reworded copy of the same message. Replace assets in batches of one or two, not all at once, so you do not force a full relearn.
4. Check the search categories against your negatives. Add negative keywords for clearly irrelevant or brand-safety query territory. Resist sculpting beyond that.
5. Review budget pacing. Use the budget reporting to project end-of-month spend and conversions before you change the daily number. Scale the winners gradually.
6. Run one structured experiment at a time. Use asset experiments to test creative concepts head to head. One variable, enough runtime to mean something, then roll the winner.

And the don't-touch list, which matters just as much:

• Do not change budgets or targets by large jumps. Gradual moves, 5 to 7 days between them.
• Do not aggressively sculpt with negative keywords. Brand safety and junk only. PMax finds profitable demand in query territory your keyword research never mapped, and over-sculpting closes those paths.
• Do not replace all your creative at once. Full creative swaps restart learning.
• Do not judge performance day by day. The algorithm optimizes over weeks. React to trends, not to Tuesdays.

Question to Answer:

Do you have a weekly PMax routine focused on insights, placements, and assets, or are you either ignoring the campaign completely or poking it every day?

12Common Performance Max Mistakes