Running Performance Max campaigns that actually convert requires a structured approach to goal definition, asset group architecture, feed quality, budget pacing, search theme strategy, and performance evaluation. Performance Max (PMax) is Google’s AI-driven campaign type that runs ads across Search, Shopping, Display, YouTube, Gmail, and Discover from a single campaign. This guide walks through seven concrete steps to set up, optimize, and scale PMax campaigns, whether you are an agency managing dozens of client accounts, an in-house team working alongside a strategist, or a business that wants the entire process handled end-to-end. By the end, you will have a repeatable PMax execution framework built for real conversions, not vanity metrics.

Before You Start

You will need a Google Ads account with active conversion tracking (ideally using GA4 and enhanced conversions), a Google Merchant Center account if running Shopping, and at least two weeks of historical conversion data in the account. If you are starting from zero, run standard Search or Shopping campaigns first to establish baseline conversion signals before launching PMax.

Have these assets ready before you begin:

  • At least 5 headlines, 5 descriptions, and 5 images per asset group
  • Your customer list or CRM data for audience signals
  • A clear understanding of your target CPA or target ROAS
  • Access to Google Merchant Center (for ecommerce or product-based businesses)

Step 1. Define Your Conversion Goal Before Touching A Single Setting

Every underperforming PMax campaign traces back to the same root cause: unclear or misconfigured goals. Before you create an asset group, choose one primary conversion action and make sure Google Ads is optimizing toward it.

Choose One Primary Conversion Action

If you sell products, your primary conversion action should almost always be purchases with revenue values passed back to Google. If you generate leads, pick your highest-quality lead event (not just form fills, but qualified submissions or booked calls). Setting multiple primary conversion actions of different types confuses PMax’s bidding algorithm and dilutes the learning signal.

Verify Your Conversion Tracking Is Passing Accurate Data

PMax is more signal-dependent than any other campaign type. If your conversion values are wrong, your tracking fires on page loads instead of confirmed actions, or your attribution window is misaligned with your sales cycle, PMax will optimize toward the wrong outcomes. Audit this before launch, not after.

Why This Matters

Unclear goals do not just waste budget. They teach PMax’s algorithm the wrong lessons, and those lessons compound over time. A campaign that spends its first two weeks optimizing toward the wrong conversion event is harder to correct than one that starts clean.

Step 2. Structure Asset Groups Around Audience Intent, Not Product Categories

An asset group in PMax functions like a mini-campaign. It combines audience signals, creative assets, and landing pages into a single targeting unit. Most advertisers structure these by product line or service category. That is a mistake. Structure them around audience intent instead.

Think Audience-First, Not Catalog-First

A skincare brand selling moisturizers, serums, and cleansers should not create three asset groups by product type. Instead, create asset groups around intent clusters: one for first-time buyers researching routines, one for brand-aware repeat purchasers, and one for ingredient-conscious shoppers comparing products. This produces creative-to-audience alignment that PMax rewards with better placements and lower CPAs.

Set Audience Signals That Are Specific Enough To Guide, Not Restrict

Audience signals in PMax are suggestions, not hard targeting. Upload your customer match lists, add custom segments based on search behavior (use real queries your buyers type), and layer in in-market and affinity audiences. Avoid broad demographics with no behavioral specificity. A “Women 25-54” signal teaches PMax nothing. “People who searched for ‘best CRM for small business’” is actionable.

How Many Asset Groups Is Optimal

For most accounts, three to five asset groups per campaign strikes the right balance. Fewer than three limits PMax’s ability to match creative to intent. More than seven typically fragments your conversion data so thinly that no asset group exits the learning phase efficiently. Agencies managing client accounts at scale through a solution like groas can test higher asset group counts because the engine processes signals across accounts faster than manual management allows, but the principle holds: start lean and expand based on data.

Step 3. Optimize Your Product Feed If You Run Shopping

For ecommerce advertisers, feed quality is the single highest-leverage input in a PMax campaign. PMax pulls product data directly from your Merchant Center feed to generate Shopping placements, and poor feed data limits which queries your products appear for and how prominently they rank.

Optimize Titles, Attributes, And Images

Your product titles should include the primary keyword a shopper would search, the brand name, and key differentiating attributes (size, color, material) in that order. Fill out every available attribute in Merchant Center: GTIN, product type, custom labels for margin-based segmentation, and shipping details. Use high-resolution images on white backgrounds as your primary image, and include lifestyle images as additional shots.

The Merchant Center Checklist Before Launching

Before you launch PMax with a Shopping feed, verify: no disapproved products, no missing GTINs on items that require them, all prices and availability matching your website in real time, and shipping and tax settings correctly configured for every region you target. One disapproved product can drag performance across the entire feed.

Common Pitfall

Advertisers often launch PMax, see mediocre Shopping performance, and assume the campaign type is the problem. In most cases, the feed is the problem. Fix your titles and attributes before adjusting bids or budgets.

Step 4. Set Your Budget And Bidding Strategy Correctly For The Learning Phase

PMax needs conversion data to learn which placements, audiences, and creative combinations work. If your budget is too low or your bidding targets are too aggressive at launch, the campaign never exits the learning phase.

Minimum Budget Thresholds For Meaningful Learning

A reasonable starting budget is at least 10 to 15 times your target CPA per day. If your target CPA is $50, budget at least $500 to $750 per day. Anything below that and PMax does not generate enough conversions to learn effectively within two weeks.

Start With Maximize Conversions Before Adding Targets

Launch with a Maximize Conversions or Maximize Conversion Value strategy without a tROAS or tCPA constraint for the first two to three weeks. This gives PMax the widest possible learning window. Once you have 30 or more conversions, layer in a tROAS or tCPA target. For a deeper breakdown of when to use target ROAS versus target CPA and the tradeoffs of each, review how each strategy affects scale differently.

How To Ramp Targets Without Triggering Resets

When you introduce a tROAS or tCPA, set it at or slightly below your actual achieved performance (not your aspirational goal). Then tighten by no more than 10 to 15 percent every two weeks. Jumping from no target to an aggressive 800% tROAS resets the learning phase and tanks delivery.

Step 5. Build A Search Theme And Negative Keyword Strategy Before You Launch

PMax will run on Search inventory whether you want it to or not. Without explicit guidance, it will absorb branded queries, compete with your existing Search campaigns, and show up for irrelevant terms that waste spend.

Set Up Brand Exclusions Explicitly

PMax does not automatically exclude branded traffic. You must request brand exclusions through your Google rep or apply them via the account-level brand list settings. If you skip this step, PMax will claim credit for conversions that your branded Search campaigns would have captured at a fraction of the cost.

Use Search Themes To Guide Intent

Search themes, introduced as a PMax feature, let you signal the types of queries you want PMax to prioritize. Add five to ten high-intent, non-branded search themes per asset group. These are not keywords in the traditional sense; they are directional signals. Use them to point PMax toward commercial and transactional queries rather than informational ones.

Build A Negative Keyword List That Protects Your Search Campaigns

Create an account-level negative keyword list targeting the branded terms and exact-match high-performers you are already capturing in dedicated Search campaigns. This prevents PMax from cannibalizing your best-performing traffic. For agencies managing multiple accounts, this step alone can prevent significant budget waste across client portfolios, and it is the type of structural safeguard that a scaled agency operation needs in place from day one.

Step 6. Evaluate Performance Using The Right Signals At The Right Time

PMax reporting is limited compared to standard campaign types. You cannot see individual keyword performance, placement-level data is aggregated, and asset performance ratings use vague labels like “Good” and “Best.” Knowing which signals to trust and when makes the difference between premature optimization and informed decisions.

Weeks 1 Through 2: Focus On Delivery And Learning Status

In the first two weeks, your primary concern is whether PMax is spending its budget and exiting the learning phase. Track impressions, spend pacing, and conversion volume. Do not make structural changes during this window unless something is fundamentally broken (disapproved assets, tracking errors, zero impressions).

After the learning phase, shift your focus to cost per conversion, conversion value per cost, and the quality of conversions (are these real customers or junk leads?). Compare PMax performance against your other campaign types on a level playing field: same attribution model, same conversion actions.

When To Rebuild Versus When To Wait

An asset group rated “Low” after three weeks with fewer than 10 conversions does not have enough data to judge. Give it more budget or consolidate it into a stronger asset group. An asset group rated “Low” after six weeks with 50+ conversions and a CPA twice your target needs a rebuild: new creative, new audience signals, potentially a different landing page. In-house teams working with a strategist through a done-with-you model benefit from having a second opinion on these rebuild decisions, because the line between patience and waste is not always obvious.

Step 7. Decide When To Let PMax Scale And When To Pull Back

PMax is not a “set it and scale it” campaign type. It requires ongoing budget allocation decisions relative to your Search campaigns, and those decisions change as your account matures.

Monitor The PMax-Search Budget Tension

PMax and Search campaigns compete for the same auction inventory. As you scale PMax budgets, watch whether your Search campaigns lose impression share, conversion volume, or efficiency. If Search performance degrades as PMax scales, PMax is likely cannibalizing high-intent traffic rather than finding incremental demand.

Watch for: declining branded Search impressions, rising PMax “conversions” with flat total account conversions, and PMax reporting strong performance on the same queries your Search campaigns already own. If you see this pattern, tighten brand exclusions, add negative keywords, and consider capping PMax budget until you isolate incremental performance.

How Each Account Type Manages This Decision

For agencies running PMax across multiple client accounts, this scaling decision multiplies across every account in the book. The automation layer underneath execution needs to flag cannibalization signals across accounts, not just within one. In-house teams with a strategist alongside them can review these patterns in biweekly strategy calls and adjust course with expert input. Fully managed accounts benefit from having a dedicated strategist who owns this tension end-to-end, making the allocation call based on full business context and real-time data, not monthly reports.

Common Mistakes To Avoid

Launching PMax with no existing conversion data. PMax needs historical signals to work. Run Search or Shopping first and accumulate at least 30 conversions before launching PMax.

Setting aggressive tROAS or tCPA targets from day one. This starves the campaign of data. Start uncapped and introduce targets gradually after the learning phase.

Creating too many asset groups with too little budget. Five asset groups splitting a $100/day budget means each asset group gets $20/day. That is not enough for any of them to learn.

Ignoring feed quality and blaming PMax. If your Shopping performance is weak, fix your titles, images, and attributes before touching campaign settings.

Failing to set brand exclusions. This is the most common and most expensive mistake in PMax. Without explicit exclusions, PMax will claim your cheapest, highest-converting branded traffic.

Evaluating PMax in isolation instead of at the account level. PMax might show great CPA while your total account conversions stay flat. Always evaluate incrementality, not just campaign-level metrics.

Never refreshing creative assets. PMax creative fatigue shows up as declining conversion rates after four to six weeks. Rotate in new headlines, descriptions, and images regularly.

How groas Handles This For You

Every step in this guide, from goal configuration and asset group architecture to feed optimization, budget pacing, search theme strategy, and scaling decisions, is execution work. It compounds across campaigns, across accounts, and across weeks. This is exactly where groas changes the equation.

For agencies running PMax across client accounts, groas provides the engine as a platform your media buyers operate directly. You connect unlimited client accounts, and the proprietary engine trained on over $500 billion in profitable ad spend handles the heavy execution underneath while your team keeps full control, their brand, and their margin. Start with a 7-day free trial and see how the engine performs across your book.

For in-house teams, groas pairs that same engine with a senior strategist who works alongside your team. You stay in control of asset strategy and business decisions while the engine manages the execution that would otherwise eat your week: bid adjustments, signal analysis, budget pacing, and performance evaluation across every PMax campaign. The strategist provides biweekly strategy calls, weekly reporting, and direct insights from groas’s team inside Google HQ. Get started with done-with-you management to keep your hands on the wheel with better tooling underneath.

For businesses that want PMax handled completely, groas as a fully managed service means a dedicated strategist owns your entire account: PMax campaign architecture, asset group strategy, feed optimization, landing pages, offer testing, and the scaling decisions covered in Step 7. Nothing to log into. Nothing to manage. Apply for DFY and groas figures out the right plan on the call.

Every product is month-to-month. No onboarding fees. No long-term contracts. groas earns the next month by performing.

Performance Max Rewards Structure, Not Luck

PMax is not a campaign type you launch and hope works. It is a campaign type that rewards precise setup, disciplined pacing, and ongoing structural decisions, the kind of execution that separates accounts that scale from accounts that stall. This guide gives you the framework. Whether you run it yourself, run it with a strategist, or hand it off entirely, the principles are the same. The difference is who does the work, and whether the execution is limited by the hours one person has in a week or powered by an engine that never stops. If you are ready to stop managing PMax by feel and start scaling it by design, groas is built for exactly that.