Performance Max Campaigns: When to Use Them and How to Optimize

Performance Max is a Google Ads campaign type that serves a single set of assets across Search, Shopping, YouTube, Display, Discover, Gmail and Maps from one budget and one bidding strategy. Instead of choosing placements and keywords, you supply inputs — a product feed, creative assets, audience signals and conversion values — and Google’s models allocate spend across inventory to hit your goal.

That trade is the whole story. You give up placement-level control and gain access to inventory that would otherwise require five separate campaigns. Whether the trade is good depends almost entirely on the quality of the inputs you can supply.

When does Performance Max fit?

The pattern is consistent: Performance Max rewards accounts that already measure well. It is not a fix for a measurement problem, and treating it as one produces a campaign that reports excellent numbers while the business sees nothing change.

The levers you actually have

You cannot bid by placement, but you are far from powerless. In practice there are seven meaningful controls.

1. The product feed. For retail, the feed is the targeting layer. Titles, product types, custom labels, GTINs, availability accuracy and image quality determine which queries you can match. A restructured feed changes performance more than any in-campaign setting.
2. Asset group structure. Each asset group has its own listing group and its own creative. Grouping products by category, margin tier or price band lets you write messaging that actually matches what is being sold.
3. Custom labels. Use them to separate high-margin from low-margin SKUs, seasonal from evergreen, or best-sellers from long tail — then build asset groups and campaigns around those labels rather than around the retailer’s internal taxonomy.
4. Conversion values. Sending gross profit instead of revenue reshapes what the bidding model chases. This is the single highest-leverage change for catalogs with uneven margins.
5. Audience signals. These are hints, not targeting. Customer lists, remarketing segments and custom segments built from search behavior accelerate the early learning phase.
6. Exclusions. Brand exclusion lists, account-level negative keywords, placement exclusions and content suitability settings keep spend away from traffic you do not want.
7. Campaign splits. Separate campaigns per country, per language or per margin tier give you independent budgets and targets. This is the closest thing to granular control that Performance Max offers.

Building the campaign properly

Step 1 — Verify measurement before anything else

Confirm that exactly one purchase or lead event fires per transaction, that the value excludes VAT and shipping if that is your convention, that enhanced conversions are enabled, and that Consent Mode is configured. If any of that is unresolved, fix it before launching.

Step 2 — Fix the feed

For each product: an accurate title that includes the terms buyers actually search, a correct product_type and google_product_category, GTIN and brand where applicable, a clean primary image, and real availability. Add custom labels for margin tier and seasonality. Resolve every disapproval in Merchant Center — suppressed products silently cap the campaign’s reach.

Step 3 — Design asset groups around meaning

One asset group per coherent group of products. Fill every asset slot: the maximum number of headlines, long headlines, descriptions, at least the required image ratios, and video. If you supply no video, the system generates one from your assets, and generated video is rarely as good as a purpose-made one — even a simple product montage outperforms it.

Step 4 — Add audience signals

Attach your customer list, site visitors segmented by depth of engagement, and a custom segment built on competitor and category search terms. Treat these as starting hints; the system will move beyond them.

Step 5 — Set a bidding target you can defend

Start with maximize conversion value without a target if the campaign is new, gather two to four weeks of data, then apply a target ROAS derived from the actual result rather than from a wish. Move targets in small increments.

Step 6 — Exclude what you do not want

Apply the brand exclusion list. Add account-level negative keywords for irrelevant categories, job seekers and free-intent queries. Set content suitability to the level your brand needs.

Reading the reports

Performance Max reporting has improved but remains thinner than standard campaigns. The places worth looking:

• Asset group performance. Compare conversion value and cost across asset groups. Chronic underperformers usually have a creative or listing-group problem, not a bidding problem.
• Search terms and categories. Available at campaign level; use it to find whole categories of irrelevant traffic worth excluding.
• Asset ratings. “Low” ratings indicate an asset the system rarely serves. Replace rather than accumulate.
• Listing group performance. Which products get impressions and which never do. A large tail of zero-impression products often points at a feed problem.
• Channel distribution. Available through scripts and reporting tooling. If nearly all spend sits in one channel, the campaign is behaving like a single-channel campaign and should be evaluated as one.

Measuring incrementality instead of taking credit at face value

The core risk with Performance Max is attribution self-flattery: it serves on the cheapest, highest-converting inventory available, including your own brand terms and returning visitors, then reports a strong ROAS for demand that already existed.

Three ways to check:

1. Baseline comparison. Record total account revenue, total spend and blended ROAS for four weeks before launch. Compare the same blended figures after. If the campaign’s ROAS is 8.0 but blended account ROAS is unchanged, the campaign is redistributing credit, not creating demand.
2. Brand isolation. Keep a dedicated brand Search campaign and apply brand exclusions to Performance Max. Watch what happens to brand campaign impression share when Performance Max scales.
3. Geographic or time-based holdouts. Where volume permits, pause the campaign in one comparable region and observe the difference in total orders rather than in platform-reported conversions.

A troubleshooting table

Performance Max alongside other campaign types

Performance Max does not have to be the whole account. A common structure keeps a dedicated brand Search campaign, a small set of exact-match Search campaigns for high-intent non-brand terms where you want explicit control, and Performance Max carrying the broad discovery and Shopping load. Demand Gen sits alongside for upper-funnel visual demand creation on YouTube, Discover and Gmail.

The point of that structure is diagnostic clarity: when something moves, you can tell which part of the account moved it.

Closing

Performance Max is neither a black box nor a magic button. It is an allocation engine whose output quality is bounded by the inputs — feed, creative, values, signals and exclusions. Accounts that treat those inputs as the real work get results. Accounts that launch it and wait get an expensive branded-search campaign with good-looking reports.

Moon Workshop manages Google Ads accounts as a Google Partner agency from Antalya, Türkiye, working with both domestic and international markets. Feed engineering, conversion value accuracy and campaign structure are handled as one system, because in Performance Max they are one system.