Google Performance Max Campaigns: Strategy, Setup & Tips

Performance Max is a single Google Ads campaign that runs across Search, Shopping, Display, YouTube, Discover, Gmail, and Maps at once, using machine learning to decide where your budget goes and which creative combination gets shown. You set the goal and supply the assets and feed data – Google handles placement and bidding. It’s powerful, but it’s also the most opaque campaign type Google offers, and that opacity is exactly where most advertisers lose money.

I’ve run Pmax across accounts ranging from small D2C stores to catalog-heavy retailers, and the campaigns that perform well share a pattern: they’re set up with intention, not just “turned on and left alone.” This guide covers what Pmax actually does, how to set one up correctly, and the 2026-specific controls – negative keywords, brand exclusions, channel reporting – that finally give advertisers a way to steer it.

What Is a Google Performance Max Campaign?

Performance Max is a goal-based Google Ads campaign type that combines Smart Bidding with automated ad placement across all of Google’s advertising channels in a single campaign. Instead of building separate campaigns for Search, Shopping, and Display, you feed Google your assets, product data, and conversion goals, and its AI decides in real time where and to whom your ads appear.

It launched in November 2021 and fully replaced Smart Shopping campaigns by mid-2023, inheriting Smart Shopping’s product feed and shopping ad functionality but wrapping it in a much broader, cross-channel automation layer.

How Performance Max Actually Works

Asset groups vs. traditional ad groups

In a standard Search campaign, you build ad groups around keywords. In Performance Max, you build asset groups instead – collections of headlines, descriptions, images, videos, logos, and a Final URL, all organized around a theme, product category, or customer segment. Each asset group functions like its own mini-campaign inside the larger Pmax structure. Google can create up to 100 asset groups per campaign, though most accounts never need anywhere close to that.

Treat each asset group as a distinct pitch to a distinct audience. An asset group mixing running shoes with winter coats gives Google’s AI conflicting signals about who should see the ads and why – the algorithm ends up guessing instead of learning.

How Smart Bidding decides where your budget goes

Performance Max only supports automated bidding – there’s no manual bid option. You choose between two focuses:

• Maximize Conversions, with an optional target CPA, when every conversion carries roughly equal value.
• Maximize Conversion Value, with an optional target ROAS, when conversions vary in value (a $15 accessory versus a $400 order, for example).

If your account has limited conversion history, starting on Maximize Conversions without a target lets the algorithm collect data quickly. Adding a tight CPA or ROAS target too early tends to choke volume before the system has learned anything useful – I’ve watched this happen on accounts that set an aggressive ROAS target in week one and then wondered why spend wouldn’t leave the ground.

Where Pmax ads can appear

Performance Max ads can show across Search, Shopping, Display, YouTube, Discover, Gmail, and Maps. The format changes automatically per channel – a Pmax ad on Gmail looks nothing like the same campaign’s ad on YouTube – which is part of why creative variety matters so much more here than in single-channel campaigns.

Performance Max vs. Standard Search and Shopping Campaigns

Performance Max isn’t a replacement for Search campaigns, even though Google often positions it that way in the interface. Search campaigns give you keyword-level control, exact match types, and full visibility into which query triggered which ad. Pmax trades that control for reach and automation.

The practical approach most experienced advertisers land on: keep a standard Search campaign running for your highest-intent, highest-value keywords – especially branded terms – and let Pmax handle broader discovery and cross-channel reach. Running Pmax as your only campaign type usually means losing granular control right when you need it most, particularly around brand protection. Google Ads management services

Is Performance Max Right for Your Business?

Performance Max rewards accounts that already have data to work with. Before launching one, check for:

• Conversion history. Pmax leans on past conversion data to figure out who to target. A brand-new account with zero conversion history gives the algorithm nothing to learn from.
• A clean, well-populated product feed (for retail) or solid landing pages (for lead gen).
• Enough budget to reach 20–30 conversions per campaign per month – Google’s general benchmark for giving the algorithm enough signal to optimize meaningfully.
• Multiple advertising goals you’d otherwise need separate campaigns to cover, like sales plus lead generation.

If you’re launching your very first Google Ads account ever, get a standard Search campaign stable first. Performance Max assumes a baseline of Google Ads fundamentals already in place.

How to Set Up a Performance Max Campaign (Step-by-Step)

Prep your Merchant Center feed

For retail Pmax campaigns, your product feed does more work than any other single input. Submit a complete feed to Google Merchant Center with accurate titles, brand, price, images, GTIN/MPN, product category, product type, and custom labels. Weak feed data caps campaign performance no matter how good your creative assets are – this is the one step advertisers most often under-invest in.

Choose campaign goal and conversion actions

In Google Ads, start a new campaign, pick an objective (Sales, Leads, etc.), and select Performance Max as the campaign type. Confirm or edit the pre-populated conversion goals – these determine exactly what the algorithm optimizes toward, so get them right before launch rather than adjusting mid-flight.

Budget and bidding strategy

Set your average daily budget and choose Maximize Conversions or Maximize Conversion Value. Decide whether to bid higher for new customer acquisition – available under Maximize Conversion Value once you have an audience segment of at least 1,000 active members to distinguish new customers from returning ones.

Building asset groups that actually convert

Use every available asset slot. More text and image variety gives the algorithm more combinations to test, which generally produces better Ad Strength and more consistent results. Google’s specs call for up to 15 headlines (15–30 characters each), at least one long headline (up to 90 characters, up to 5 total), 2–5 descriptions (up to 90 characters each), and a business name (up to 25 characters). For images, provide a landscape marketing image (1200×628 recommended), a square marketing image (1200×1200 recommended), and a logo (1200×1200 recommended, square aspect ratio) at minimum.

Skip uploading your product feed images as separate assets – that just creates duplicates Google already has access to. Instead, use the image asset slots for lifestyle shots: real context, real people where appropriate, one clear focal point taking up roughly 40% of the frame, and no text overlay.

Audience signals – what to include and what’s a waste of time

Audience signals aren’t hard targeting – they’re a starting point that tells the algorithm who’s likely to convert. Include remarketing lists, Customer Match data, and relevant in-market or affinity segments where you have them. Skip guessing at vague demographic combinations with no data behind them; the algorithm will find its own audience beyond your signals anyway once it has enough conversion volume to work with.

Location and final URL expansion settings

Choose “Presence” rather than the default “Presence or interest” location targeting if you only want to reach people physically inside your target area – the default also serves ads to people merely interested in that location, which dilutes relevance for most local or regional advertisers.

For Final URL Expansion, decide whether Google can redirect traffic to pages it judges more relevant than your specified URL. Turning it off keeps full control over destination pages; leaving it on can find better-converting landing pages you hadn’t considered, at the cost of some visibility into where traffic lands. Exclude non-commercial URLs (FAQ, About, shipping policy) either way, since they rarely convert.

2026 Performance Max Updates Advertisers Need to Know

Pmax has changed enough since 2023 that guidance written even a year ago is already outdated in places. Here’s what’s actually different now.

Campaign-level and account-level negative keywords

For years, advertisers couldn’t add negative keywords to Pmax directly – you had to file a request through Google support and wait. That’s no longer true. Campaign-level negative keywords are now fully supported, with a limit of up to 10,000 per campaign, and they behave the same way negative match types do in standard Search campaigns. Account-level negative keyword lists continue to work alongside them, applying exclusions across Search, Shopping, and Pmax simultaneously.

This matters because it closes the single biggest control gap Pmax had since launch. If you’ve been avoiding Pmax because of the “black box” search query problem, that objection is largely outdated now.

Brand exclusion controls

Google now recommends using dedicated brand exclusion controls, rather than plain negative keywords, when the goal is blocking your own or a competitor’s branded terms. Brand exclusions are more precise for this specific use case and less likely to accidentally block a legitimate non-branded query that happens to share a word with a brand name.

Channel-level reporting and the expanded Search Terms report

Reporting transparency has genuinely improved. The Insights tab and channel-level reporting now show budget allocation across Search, Shopping, Display, YouTube, Discover, Gmail, and Maps, rather than treating the campaign as one undifferentiated black box. The Search Terms report, first introduced in a limited form in 2023, gives visibility into which queries actually triggered impressions – still not as granular as a standard Search campaign’s report, but a real improvement over the early years of Pmax.

AI Max overlap and signal conflicts

Google’s newer AI Max features for Search campaigns and Performance Max increasingly share underlying signal infrastructure, and the two don’t always play cleanly together in the same account. If you’re running AI Max on Search alongside Pmax, watch for overlapping targeting signals pulling from the same audience pools – this is an evolving area, and Google’s own documentation on how the two interact keeps shifting, so treat any specific interaction advice as provisional and verify against current settings before making structural changes.

Performance Max Optimization Strategies That Move ROAS

Structuring campaigns by margin, not just category

Don’t combine high-margin and low-margin products in the same Pmax campaign. The algorithm optimizes toward whatever converts most efficiently, and lower-margin, higher-converting SKUs will pull budget away from higher-margin products that convert less often but matter more to your bottom line. If your catalog spans both, split them into separate campaigns with separate goals.

Feed optimization as the real lever

Title structure, in particular, deserves attention: Google weighs the first few words in a product title heavily when matching ads to search queries, so lead with the most identifying information (brand, product type, key attribute) rather than burying it after filler words. Fill gaps in weak attributes by extracting data from other fields where possible, and test alternate title structures against each other over time rather than guessing once and leaving it.

Asset performance grading and refresh cadence

Check the Asset Detail report regularly. Assets get graded Low, Good, Best, or Pending (once they cross roughly 5,000 impressions). Replace Low-performing assets rather than letting them sit – they’re actively dragging down the combinations Google tests. Don’t confuse individual asset performance with the overall Ad Strength score; Ad Strength reflects asset variety and volume, not results, so a “poor” Ad Strength label doesn’t automatically mean poor performance.

Using audience signals without over-constraining the algorithm

Revisit audience signals every few months rather than setting them once at launch. As your remarketing lists and Customer Match data grow, feeding fresher signals back into the campaign tends to sharpen targeting over time – but avoid stacking so many narrow signals that you’re effectively hand-picking an audience, which undercuts the point of letting the algorithm find incremental reach.

When and how to exclude branded search

Exclude your own branded terms from a general Pmax campaign and run brand terms through a dedicated Search campaign instead. Branded traffic converts at a different rate and typically deserves a different ROAS target than non-branded traffic – folding both into one Pmax campaign can make overall performance look artificially strong while actually starving new customer acquisition, since the algorithm gravitates toward whatever converts best, and branded search usually wins that comparison easily.

Common Performance Max Mistakes to Avoid

• Launching with a tight ROAS target before the account has conversion history. Give the algorithm room to learn first.
• Mixing conversion goals in one campaign – e-commerce sales and lead-gen form fills pulling in different directions confuses optimization.
• Reusing product feed images as separate image assets, creating redundant creative instead of new variety.
• Ignoring the Search Terms report now that it exists – the visibility gap that used to excuse this is mostly closed.
• Judging results too early. Pmax campaigns typically need several weeks of stable spend before performance data means much; judging week one against a mature Search campaign isn’t a fair comparison.

Performance Max Pros and Cons

What to Do When Performance Max Underperforms

Before assuming Pmax itself is the problem, check three things in order: conversion tracking accuracy, feed data quality, and whether the campaign has had enough time and volume to exit its learning phase. In my experience, a surprising share of “Pmax doesn’t work for us” cases trace back to broken or incomplete conversion tracking rather than the campaign structure – the algorithm optimized correctly against bad data, and bad data produces bad results regardless of how well the campaign is built.

If tracking and feed data check out and performance still lags, look at whether branded and non-branded traffic are mixed together, whether asset groups are too broad, and whether the Search Terms report shows the campaign matching to genuinely irrelevant queries. Google Ads audit service

FAQ

Is Google Performance Max better than Smart Shopping? 

Smart Shopping no longer exists – Google fully migrated all Smart Shopping campaigns to Performance Max by mid-2023. Pmax kept the shopping ad functionality Smart Shopping had and added cross-channel reach across Search, Display, YouTube, Discover, Gmail, and Maps, along with generative AI asset creation.

How long should you run a Performance Max campaign before judging results? 

Most practitioners recommend at least four to six weeks of stable spend before drawing conclusions, since the algorithm needs a learning period to gather conversion data and stabilize bidding.

Can you add negative keywords to Performance Max campaigns? 

Yes. As of 2026, campaign-level negative keywords are fully supported, with a limit of up to 10,000 per campaign, alongside account-level negative keyword lists that apply across Search, Shopping, and Pmax together.

How much budget do you need to run Performance Max effectively? 

There’s no fixed minimum, but Google generally recommends enough budget to reach 20–30 conversions per campaign per month so the algorithm has sufficient signal to optimize against.

Does Performance Max cannibalize other campaigns? 

It can. Pmax tends to prioritize whatever converts most efficiently, which is often branded search traffic already captured by existing Search campaigns. Excluding branded terms from Pmax and running them through a dedicated Search campaign helps limit this overlap.

Should you run Performance Max alongside Search campaigns? 

For most advertisers with an established branded search presence, yes. Keep a standard Search campaign for high-intent branded and category keywords, and let Pmax handle broader cross-channel discovery rather than relying on it exclusively.

Conclusion

Performance Max works best as one part of a broader Google Ads structure, not a replacement for everything else you’re running. The advertisers who get real value out of it treat the automation as a starting point that needs feeding, watching, and periodic correction – clean conversion data, a well-built feed, varied creative, and now, negative keywords and brand exclusions to keep it from wandering into irrelevant territory. The 2026 controls have closed most of the gap that made Pmax frustrating to manage; what’s left is mostly a matter of setting it up with the same discipline you’d apply to any other campaign type, then giving it enough time and data to actually do its job.