On this page
Performance Max runs across Google's seven channels from a single campaign, but for ecommerce brands, Shopping placements drive the majority of conversions. Those placements run entirely on Merchant Center product data. Not on your headlines. Not on your audience signals. On the feed. Every guide tells you to clean up your feed before launching PMax. None of them explain what actually breaks when you don't. When the feed is incomplete, the algorithm doesn't just show fewer products. It routes spend into channels with weaker purchase intent, burns budget on queries that don't match your catalog, and generates a learning-period signal that teaches the campaign the wrong patterns from day one.
This guide covers the feed setup sequence Hellyeah uses when onboarding ecommerce brands onto Performance Max, the specific attributes that determine Shopping placement eligibility, and the one benchmark that tells you whether your feed is limiting your campaign before you touch a single bid setting.
Quick Answer
- The Performance Max product feed is the query-matching layer for Shopping placements. Without complete product data, PMax cannot show your products for relevant searches regardless of how the campaign is configured.
- Feed errors that prevent Shopping impressions: missing GTINs on branded products, generic product titles that don't match buyer search terms, images below 100x100 px, and Merchant Center disapprovals from price or availability mismatches.
- A healthy ecommerce PMax account routes 60-80% of its budget through Shopping placements. If your Shopping share is below 50%, the feed is the first thing to investigate, not the campaign settings.
The Performance Max product feed is the query-matching layer that determines which Shopping placements your campaign is eligible for. Google uses product titles, categories, GTINs, and custom attributes to match user searches to specific products in your Merchant Center catalog. When those attributes are missing or generic, the algorithm cannot match your inventory to the queries that would convert. Hellyeah's AIMA engine handles bidding, audience signals, and cross-channel budget allocation for PMax campaigns, and AIMA can only optimize within the inventory that the feed makes eligible. A product that doesn't appear in Shopping because its title is too generic or its GTIN is missing never generates a conversion signal for AIMA to learn from. Feed quality is not a prerequisite for launching PMax. It is a prerequisite for any AI optimization system having anything useful to work with.
Why Feed Quality Is the Only Ceiling Performance Max Cannot Lift
Performance Max campaigns use machine learning to allocate budget across Search, Shopping, YouTube, Display, Gmail, and Discover simultaneously. For most ecommerce accounts, Shopping ends up as the dominant placement by conversion volume. But unlike Search, which can generate impressions from any query that matches a keyword pattern, Shopping placements require a product to be eligible. Eligibility is determined entirely by the Merchant Center feed.
What the Feed Controls in a PMax Campaign
When a user searches "blue trail running shoes size 10," Google matches that query against products in Merchant Center. The match depends on the product title, product type, category, and attributes like size and color. PMax has no keyword targeting. The feed is the targeting. If your product title is "Men's Running Shoe, Model 74B" with no color or size data and a vague product type of "Footwear," the algorithm cannot confidently serve your product for that high-intent query. It will serve a competitor whose title reads "Men's Blue Trail Running Shoe, Size 10, Wide Fit."
How Feed Completeness Limits AI Bid Optimization
Hellyeah's AIMA engine adjusts bids and audiences in real time based on conversion signals from across the campaign. When Shopping placements are suppressed because the feed is incomplete, AIMA has less conversion data to work from. A campaign running at 30% Shopping share because half the catalog is disapproved gives AIMA a distorted signal: it appears to show that Shopping underperforms, when the reality is that only the weakest products are showing there. The AI then routes more budget toward non-Shopping channels, not because they perform better, but because Shopping has been starved of its best inventory. Fix the feed and the Shopping share recovers, typically within 7 to 14 days as the algorithm relearns with clean data.
In a live campaign, the simplest version of this pattern shows up as high Display spend and low Shopping spend for an account that should convert heavily through Shopping. The algorithm didn't choose that allocation based on performance. It defaulted to it because the Shopping inventory was too thin.
What to Fix Before Your PMax Campaign Starts Learning
The learning period for a new PMax campaign typically runs two to four weeks. During that window, the algorithm tests placements and allocates budget toward signals that predict conversion. If the feed is broken during that period, the campaign learns from bad data. A model trained on incomplete inventory is harder to correct than one that was never trained in the first place. Run this feed gate before launch, not after.
Resolve Merchant Center Disapprovals First
A disapproved product does not show in Shopping placements or free listings. PMax pulls from the same Merchant Center data, so a disapproved product is invisible to the campaign regardless of how the asset group is configured. The most common causes of disapproval:
- Price mismatch between the feed and the landing page (the feed says $49.99, the product page says $54.99)
- Availability marked as "in stock" in the feed but "sold out" on the product page
- Missing required attributes for product category (GTINs for branded goods, MPNs for manufacturer parts)
- Image policy violations (promotional overlays, watermarks, or insufficient resolution)
Open Merchant Center Diagnostics before touching any campaign setting. Products in "disapproved" or "limited" status are the first thing to resolve. A catalog with 15% of products disapproved has 15% of its potential Shopping inventory offline.
Check Image Minimum Requirements
Google's minimum image resolution for non-apparel products is 100x100 pixels; for apparel, it's 250x250 pixels. In practice, those minimums produce low-quality Shopping ads. Hellyeah recommends a minimum of 800x800 pixels for product images entering PMax, with white or neutral backgrounds for catalog images. Lifestyle images can be added as supplemental image links once the catalog images pass validation.
The practical issue for most ecommerce catalogs: product images were sized for website thumbnails, not ad placements. A product image at 300x300 pixels passes Merchant Center validation but competes poorly in Shopping carousels against competitors using 1200x1200 pixel images with multiple angles. Image quality is a visible component of Shopping ad CTR, affecting which product a buyer clicks before any bid optimization has a chance to matter.
Cover GTINs and Product Identifiers
GTINs (Global Trade Item Numbers, the product's barcode) tell Google's systems that a specific product in your feed matches a product cataloged across the web. When a product has a GTIN, Google can cross-reference it against price comparison data, reviews, and competitive bids from other advertisers. Products with verified GTINs generally earn better impression share at lower bids than products with identifier_exists: false.
For branded products (anything with a manufacturer name and a manufacturer barcode), include the GTIN. For private-label or custom products with no standard identifier, set identifier_exists: false and fill in the brand, MPN, and detailed product type attributes instead. Omitting the GTIN on a branded product is one of the fastest ways to suppress Shopping impression share without generating a disapproval. The product shows, just less often and at higher CPCs.
How to Write Product Titles That Win Shopping Placements
Product titles are the single highest-impact feed attribute. Google's Shopping matching algorithm uses the title as its primary signal for query relevance. A weak title means fewer impressions and lower CTR. A well-structured title expands the query surface your product is eligible for and aligns with how buyers actually search.
Title Structure That Matches How Buyers Search
The standard title structure for Shopping has been documented across enough accounts to be treated as a baseline: Brand + Product Type + Key Differentiating Attribute + Secondary Attribute. For a performance apparel brand, that produces titles like "Nike Men's Trail Running Jacket, Waterproof, Large" rather than "Men's Jacket NJ-42."
Practical rules that hold across ecommerce categories:
- Keep titles between 70 and 150 characters. Below 70, you're leaving relevant attribute coverage on the table. Above 150, Google truncates the title in Shopping carousels and stops scoring the rest for matching.
- Lead with brand name for branded products. For private-label goods, lead with product type. Buyers searching "Patagonia fleece jacket" and buyers searching "fleece jacket men's" are different queries requiring different title structures.
- Use digits, not words. "32GB" beats "thirty-two gigabytes" for matching and legibility.
- Put color, size, and material early in the title when those are the primary differentiators between SKUs. Buyers filter by these attributes and the algorithm uses them for matching.
Common Title Patterns That Suppress Impressions
The most common title problems Hellyeah sees when onboarding ecommerce brands for PMax campaigns: internal product codes embedded in titles, vague product type descriptors ("Item," "Product," "Goods"), marketing language in the title field ("Best," "Sale," "Amazing Quality"), and missing size and color when the product comes in multiple variants.
Variant products are a specific problem. If a catalog has a product in 12 colors and 6 sizes, every SKU needs its own title that specifies color and size. A title like "Women's Running Shorts" covers all 72 SKUs but matches no specific query as well as "Women's Running Shorts, Navy, Small." Feed rules inside Merchant Center can automate title construction from variant attributes without requiring a manual edit for every SKU.
How Custom Labels Give AIMA More Budget Precision
Custom labels are feed attributes with no standard Google definition. You assign the values and use them for campaign segmentation. They're the mechanism for telling PMax how to group and treat different parts of your catalog. Without custom labels, the algorithm manages all products in an asset group identically, regardless of margin, seasonality, or conversion history.
Label by Margin Tier
The most useful custom label for ecommerce PMax campaigns is a margin tier: high-margin, mid-margin, low-margin. A high-margin product that converts well can sustain a higher target return on ad spend, or approached from the other side, can absorb a lower ROAS target to drive volume while still meeting profitability requirements. A low-margin product needs a tighter ROAS to remain profitable, which in PMax means a more restrictive target that may limit impression volume.
Setting custom labels by margin tier lets Hellyeah create separate asset groups for each tier within a PMax campaign, each with its own ROAS target calibrated to profitability, rather than one blended ROAS target that underinvests in high-margin products and overinvests in low-margin ones.
Use Labels to Protect High-Performers from Low-Value Inventory
Excluding underperforming products is one of the clearest feed interventions in PMax. Products with zero conversions after 90 days, products with CPCs above the margin threshold, and products with high return rates all draw budget that would perform better on products with proven conversion histories.
Custom label the bottom 20% of your catalog by conversion value, then exclude those products from your primary PMax asset group. Run them in a separate, lower-budget asset group if you need to maintain visibility. This doesn't starve the algorithm of data. It focuses the algorithm's learning on inventory that can actually convert at a profitable cost, which produces a cleaner signal for budget optimization across the account.
Author's Comment
"The product question we kept coming back to when building the PMax setup flow inside Hellyeah was: what does the ecommerce brand actually control here, and what should the platform handle for them? The feed is the one thing the brand controls that the platform cannot substitute. AIMA can move bids in real time, rebalance budget across channels, and suppress underperforming creative, but it cannot fix a product title or resolve a Merchant Center disapproval. The sequence matters: feed first, then let AIMA optimize the campaign around clean inventory."
Yulong He, Product, Hellyeah AI
What Does the Shopping Budget Split Tell You About Feed Health?
A healthy Performance Max campaign for an ecommerce brand routes 60-80% of its budget through Shopping placements. If your Shopping budget share is below 50% on a campaign where Shopping conversions should dominate, the feed is the first thing to investigate, not the campaign settings, not the asset quality, and not the bid strategy. Below-50% Shopping share in a purchase-optimized PMax campaign typically indicates one of three feed conditions: a high disapproval rate suppressing Shopping-eligible inventory, product titles that are too generic to match high-intent queries, or a catalog structure where low-value products are cannibalizing budget from higher-converting SKUs. Hellyeah's X-Ray shows channel-level spend allocation for PMax, making this diagnostic a 30-second check rather than a multi-report pull.
Pull the channel performance breakdown from your Google Ads account (Campaigns > Performance Max campaign > Insights tab > Asset Group Performance, or Channel Performance if your account shows it). The data shows how budget has been allocated across Shopping, Search, YouTube, Display, and Gmail over the period you select.
Practitioner benchmarks for a purchase-optimized ecommerce PMax campaign:
- 60-80% Shopping: healthy for most ecommerce catalogs
- 40-60% Shopping: investigate feed quality and product exclusion settings
- Below 40% Shopping: feed issues, conversion tracking problems, or the campaign is optimizing for a non-purchase goal
The Shopping share is a lagging indicator. It reflects where the algorithm found conversions over the past 7-14 days. A feed fix today won't show in the Shopping share until after the algorithm has had time to discover and serve the newly eligible products. For conversion lift testing in Performance Max, a stable Shopping budget share is a prerequisite. Campaigns with unstable channel splits produce inconclusive lift results because the treatment and control groups are being served different channel mixes.
How Hellyeah Sets Up PMax for Ecommerce Brands
When an ecommerce brand connects Hellyeah for PMax management, the onboarding sequence follows a feed-first order before any campaign is launched. The sequence protects the learning period from being corrupted by feed issues that could have been caught before launch.
From Merchant Center to Live PMax
- Merchant Center account connected and verified, with products linked to the Hellyeah workspace
- Feed diagnostic run: disapproval rate, title quality score, image compliance, GTIN coverage
- Critical feed issues resolved or flagged for the brand to fix (Hellyeah handles campaign-side configuration; the ecommerce platform or feed management tool handles product data edits)
- Custom labels applied for margin tier and product exclusions
- Asset group structure built around the labeled catalog, one group for high-margin products, one for mid-tier, exclusions applied to the bottom performers
- PMax campaign launched in feed-only mode (no additional creative assets in the initial asset group), giving the algorithm a clean Shopping signal before Display or YouTube placements are introduced
- AIMA begins optimization with Maximize Conversion Value on day one, shifting to a target ROAS once the campaign has accumulated 30 conversions
The feed-only launch is a deliberate choice. Adding lifestyle images, headlines, and video assets to a PMax asset group opens up YouTube and Display inventory. That's valuable eventually, but in the first two to four weeks, the priority is for the algorithm to learn which products convert through Shopping at what bid, without the noise of Display and YouTube conversion patterns mixing into the signal. Once Shopping performance is established, Hellyeah adds the full creative asset set to activate the broader inventory.
What AIMA Handles and What the Feed Must Supply
Hellyeah's AIMA engine manages bids, audiences, budget pacing, and creative rotation. The feed supplies product eligibility, query matching, and the product-level signals that AIMA uses to decide which items to emphasize. These are not interchangeable.
AIMA cannot compensate for a feed where 20% of products are disapproved. It cannot improve Shopping impression share by changing a bid. It cannot construct a product title that matches a buyer's search query. What AIMA does with a clean feed: it identifies which products in a high-margin segment are converting at a profitable cost, shifts budget toward those items faster than a human optimizer could, and suppresses spend on items that are drawing clicks without converting, all within the inventory that the feed has made eligible.
For AI marketing for ecommerce brands, the pattern holds across platforms: AI optimization compounds with clean data and degrades with messy data. The feed is not a technical prerequisite. It's an investment in the quality of every decision AIMA makes downstream.
For ecommerce brands running PMax without a platform like Hellyeah, the same principles apply. The automation handles less, which means manual paid ads management hits its ceiling past $50K in ways that feed-level issues compound further.
Frequently Asked Questions about Performance Max Product Feeds
How many products should I include in a Performance Max campaign?
There's no upper limit, but the practical constraint is data quality. A catalog of 10,000 SKUs where 3,000 are disapproved and 4,000 have generic titles is less useful than a catalog of 2,000 well-optimized products. Start with your top 50 to 100 products by conversion history, optimize those feeds first, and expand to the full catalog once the high-performers are clean and the campaign has established a Shopping baseline. Hellyeah structures this as a separate high-performance asset group rather than a filtered campaign.
Does Performance Max need a Merchant Center feed to show Shopping ads?
Yes. Without a Merchant Center product feed linked to the PMax campaign, Shopping placements are not available. The campaign will still run across Search, Display, YouTube, Gmail, and Discover using your asset group creative, but Shopping requires product-level data from Merchant Center. Hellyeah's onboarding confirms the Merchant Center connection before any PMax campaign is created.
How long does it take for feed changes to show up in PMax Shopping performance?
Merchant Center typically processes feed updates within 24 to 48 hours. After a feed fix, such as resolving disapprovals, updating titles, or adding GTINs, allow 7 to 14 days for PMax to discover and test the newly eligible products before evaluating performance. Evaluating a feed fix after 48 hours is too early; the algorithm needs time to route impressions to the updated inventory.
Should I use product-level bidding or campaign-level bidding in PMax?
PMax doesn't support product-level bidding directly. The equivalent is asset group segmentation with different ROAS targets per group. Hellyeah uses custom labels to separate high-margin and mid-margin products into different asset groups, each with a ROAS target calibrated to the margin available. This produces the budget precision of product-level bidding within PMax's campaign architecture.
What is a feed-only Performance Max campaign?
A feed-only PMax campaign is one where the asset group contains product feed data from Merchant Center but no additional creative assets, meaning no headlines, descriptions, images, or videos beyond what the feed provides. Google serves this campaign primarily through Shopping placements, with limited exposure on other channels. This configuration is useful during the learning phase when you want clean Shopping data before Display and YouTube inventory are introduced. Hellyeah uses feed-only asset groups for the first two to four weeks of a new ecommerce PMax campaign.
How do I know if my PMax campaign is spending on Shopping vs. other channels?
The channel performance data is in the Google Ads interface under your PMax campaign. Check Insights > Channel Performance, or segment by network in the campaigns view. Hellyeah's X-Ray pulls this data automatically and flags campaigns where Shopping spend falls below the 60% threshold for purchase-optimized ecommerce accounts. A Shopping share below that range in a conversion-optimized account typically signals a feed or conversion tracking issue rather than a campaign configuration problem.
Can I run Performance Max and Standard Shopping campaigns at the same time?
Yes. Since October 2024, Google uses Ad Rank rather than campaign type to determine auction priority. Running both simultaneously is a valid structure, particularly for large catalogs where Standard Shopping gives you tighter product-level control over top performers while PMax handles broader inventory. Hellyeah evaluates the hybrid approach on a case-by-case basis depending on catalog size, conversion volume, and how tightly the brand wants to control query-level spend.
What feed attributes matter most for Performance Max Shopping performance?
In order of impact: product title (query matching), GTIN and product identifier (impression share and competitive positioning), product category (matching quality), images (CTR once the impression is earned), and custom labels (campaign segmentation and bidding control). Price accuracy and availability affect disapproval rate and therefore eligibility. Everything else, such as descriptions, product highlights, and additional images, improves performance at the margin but doesn't change fundamental eligibility.
Related blog posts
Frequently asked questions
How many products should I include in a Performance Max campaign?
There is no upper limit, but the practical constraint is data quality. Start with your top 50 to 100 products by conversion history, optimize those feeds first, and expand to the full catalog once the high-performers are clean and the campaign has established a Shopping baseline.
Does Performance Max need a Merchant Center feed to show Shopping ads?
Yes. Without a Merchant Center product feed linked to the PMax campaign, Shopping placements are not available. The campaign will still run across Search, Display, YouTube, Gmail, and Discover using your asset group creative, but Shopping requires product-level data from Merchant Center.
How long does it take for feed changes to show up in PMax Shopping performance?
Merchant Center typically processes feed updates within 24 to 48 hours. After a feed fix, allow 7 to 14 days for PMax to discover and test the newly eligible products before evaluating performance.
Should I use product-level bidding or campaign-level bidding in PMax?
PMax does not support product-level bidding directly. The equivalent is asset group segmentation with different ROAS targets per group. Use custom labels to separate high-margin and mid-margin products into different asset groups, each with a ROAS target calibrated to the margin available.
What is a feed-only Performance Max campaign?
A feed-only PMax campaign is one where the asset group contains product feed data from Merchant Center but no additional creative assets. Google serves this campaign primarily through Shopping placements. This configuration is useful during the learning phase when you want clean Shopping data before Display and YouTube inventory are introduced.
How do I know if my PMax campaign is spending on Shopping vs. other channels?
Check the channel performance data in Google Ads under your PMax campaign: Insights > Channel Performance, or segment by network in the campaigns view. A Shopping share below 60% in a conversion-optimized ecommerce account typically signals a feed or conversion tracking issue.
Can I run Performance Max and Standard Shopping campaigns at the same time?
Yes. Since October 2024, Google uses Ad Rank rather than campaign type to determine auction priority. Running both simultaneously is a valid structure, particularly for large catalogs where Standard Shopping gives tighter product-level control over top performers while PMax handles broader inventory.
What feed attributes matter most for Performance Max Shopping performance?
In order of impact: product title (query matching), GTIN and product identifier (impression share and competitive positioning), product category (matching quality), images (CTR once the impression is earned), and custom labels (campaign segmentation and bidding control). Price accuracy and availability affect disapproval rate and therefore eligibility.

Product
Product at Hellyeah. Designs the surface where people and automation share a workspace: what to expose, what to automate, what to leave alone.




