What is Google Performance Max and how does it work?
Performance Max is a conversion-focused campaign that runs across all of Google's inventory from one campaign. Instead of building separate Shopping, Search, YouTube, and Display campaigns, you give Google a product feed, creative assets, a budget, and a goal, and it competes for every eligible auction across its network.
It works on Smart Bidding. You pick a conversion goal, then either Maximize Conversions (for volume) or Maximize Conversion Value (for revenue), with an optional Target ROAS (tROAS) to hold a value threshold. Google predicts the value of each impression and bids automatically, so the quality of your conversion tracking directly determines the quality of the results.
For a store, the product feed in Google Merchant Center is the engine. PMax reads titles, images, prices, and availability from your feed and builds Shopping-style ads automatically, then supplements them with your uploaded text, image, and video assets to fill Display, YouTube, and Discover placements.
- Serves on eight surfaces: Search, Shopping, YouTube, Display, Gmail, Discover, Maps, and Search partners
- Requires a linked Google Merchant Center feed for product ads
- Runs on Maximize Conversions or Maximize Conversion Value, with optional Target ROAS or Target CPA
- Learns from your conversion data — thin or misconfigured tracking produces weak bidding
- Enters a ~2-6 week learning phase after launch or after major edits
What are asset groups in Performance Max?
An asset group is a themed bundle of creative — headlines, descriptions, images, logos, videos — combined with a set of products (a listing group) and an audience signal. It is the closest thing PMax has to an ad group, and it is how you segment a campaign.
Google mixes and matches assets within a group to build ads for each placement, so you should theme each asset group around one product category, collection, or margin tier. One asset group selling running shoes should not also carry copy about kitchen knives.
Give Google enough raw material to work with. Uploading your own images and at least one video matters — if you leave video blank, Google auto-generates a low-effort slideshow from your images that often underperforms and cheapens the brand.
- Text: up to 5 short headlines (30 char), 5 long headlines (90 char), 5 descriptions (90 char), 1 business name
- Images: multiple ratios — 1.91:1 landscape, 1:1 square, and 4:5 portrait; upload at least a few of each
- Logo: 1:1 square and 4:1 landscape versions
- Video: 10+ seconds; add your own or Google auto-creates one
- Structure: one asset group per category or theme, not one giant catch-all group
What are audience signals and do they matter?
Audience signals are suggestions you give Google about who is likely to convert — not hard targeting. Google uses them to accelerate the learning phase, then expands beyond them if it finds converting users elsewhere. This is the single most misunderstood control in PMax.
The strongest signals are your first-party data: customer match lists (past purchasers, email subscribers) and website visitors from your remarketing tags. Layer in custom segments built from high-intent search terms, competitor URLs, and relevant apps, then add interests and demographics as a weaker third tier.
Because signals guide rather than restrict, do not expect PMax to stay inside them. Their job is to shorten learning and improve early efficiency, so feed the algorithm your best-converting audiences even though it will eventually prospect wider.
- Your data (customer match, site visitors) — strongest signal
- Custom segments — people who searched specific terms or browsed specific sites/apps
- Interests and demographics — broad, use as a supporting layer
- Signals speed learning; they do not cap reach — Google can and will go beyond them
When should you use Performance Max vs Standard Shopping?
Use Performance Max when you want full-funnel reach and scale, have solid conversion tracking (ideally with real conversion values), and are comfortable trading control for automation. It shines for stores with a broad catalog, a reasonable creative library, and enough conversion volume for the algorithm to learn — a practical floor is roughly 30+ conversions in the trailing 30 days.
Use Standard Shopping when you need control: negative keywords, search-term visibility, granular product bidding, or the ability to protect a thin catalog from being buried in auto-generated Display placements. Standard Shopping is also better when you are diagnosing feed or bidding issues, because it is transparent in ways PMax is not.
Know the priority rule: if a PMax campaign and a Standard Shopping campaign target the same products in the same account, PMax wins the auction. You cannot cleanly A/B them on the same SKUs — PMax will simply outrank Standard Shopping, so segment products between them if you run both.
- Choose PMax for: scale, full-funnel reach, strong conversion data, broad catalog
- Choose Standard Shopping for: control, transparency, small catalogs, diagnostic work
- PMax outranks Standard Shopping for the same products — split SKUs if running both
- A common hybrid: PMax for the main catalog, Standard Shopping for a niche or high-margin subset you want tight control over
What are the biggest Performance Max pitfalls for e-commerce?
Brand traffic cannibalization is the number one hidden problem. Left unchecked, PMax happily serves ads to people already searching your brand name — cheap clicks that would have converted anyway — which inflates the campaign's ROAS and makes it look far better than the incremental revenue it actually drives.
The fix is brand exclusions. Apply an account-level brand list to exclude your own brand from the PMax campaign, and run branded search separately in a dedicated Search campaign where you can see and control it. This forces PMax to earn genuinely new demand and gives you an honest read on performance.
The second pitfall is lack of control and transparency. PMax historically hid search terms, placements, and channel-level spend, so you cannot easily see whether budget went to a valuable Search auction or a junk Display placement. Related traps: PMax cannibalizing your existing Search campaigns, a weak or unoptimized product feed starving the algorithm, and launching with too little conversion data for the learning phase to stabilize.
- Brand cannibalization: exclude your brand via an account-level brand list; run branded Search separately
- Low transparency: use PMax placement, search-term insight, and third-party scripts to surface where spend goes
- Feed neglect: a thin feed (bad titles, missing GTINs, low-res images) caps performance no matter the budget
- Data starvation: don't launch tROAS with too few conversions — start on Maximize Conversion Value, add tROAS once you have volume
- Creative laziness: skip Google's auto-generated video and upload your own to control brand and quality
How do you set up a Performance Max campaign? A step-by-step checklist
Setup determines 80% of the outcome. Get tracking, feed, and structure right before you touch budget or bids, because PMax amplifies whatever data foundation you give it.
Work through this in order:
- 1. Verify conversion tracking — purchases fire with accurate dynamic values; use enhanced conversions and, ideally, mark purchase as the single primary conversion action
- 2. Optimize the Merchant Center feed — accurate titles with keywords front-loaded, GTINs, high-res images, correct pricing and availability, and product-level custom labels for later segmentation
- 3. Choose the goal and bidding — start on Maximize Conversion Value; add a Target ROAS only once you have ~30+ conversions of history to anchor it
- 4. Set the budget — allow at least 3x your target CPA per day so the campaign can exit learning without starving
- 5. Build asset groups by theme — one per category or margin tier, each with full text, multiple image ratios, logos, and a real video
- 6. Add audience signals — lead with customer match and site visitors, then custom segments from high-intent search terms
- 7. Apply brand exclusions — add an account-level brand list to keep PMax off your own branded searches
- 8. Set final-URL expansion and exclusions — control which landing pages Google can send traffic to
- 9. Turn on the campaign and freeze it — avoid major edits for 2-3 weeks so the learning phase can complete
How do you optimize Performance Max after launch?
Optimization is patient and data-led, not daily knob-twiddling. Every significant edit can reset learning, so batch changes and give each one time to prove out over at least a week or two of conversion data.
Focus on the levers that actually move PMax: creative quality, feed quality, product segmentation, and conversion-value accuracy. The algorithm does the bidding — your job is to improve the inputs it bids on.
Track incrementality, not just reported ROAS. After adding brand exclusions, expect the headline ROAS to drop while true incremental revenue holds or improves; that lower-but-honest number is the one to optimize against.
- Review the Asset group reporting and replace assets rated 'Low' — aim for 'Good' or 'Best' on headlines, images, and video
- Watch the learning phase — don't judge or heavily edit the campaign until it stabilizes (~2-6 weeks)
- Segment by performance — split winning vs losing products into separate campaigns or asset groups using custom labels, so you can set different tROAS targets
- Adjust Target ROAS in small steps (5-10%) — jumping too high too fast collapses volume
- Use search-term insights and the placement report (or Google Ads scripts) to spot brand cannibalization and junk placements
- Fix the feed continuously — better titles and images lift Shopping performance more than any bid change
- Compare against a pre-PMax baseline or run a lift test to confirm incremental, not just reported, gains