Google Shopping Campaigns: Why Bids Are Not the Problem

Most brands optimize their Google Shopping campaigns from the wrong end. They adjust bids, test Performance Max structures, and reshuffle ad groups. ROAS drifts anyway.

Google Shopping runs on three layers: product data at the bottom, campaign architecture in the middle, and bid strategy at the top. Nearly every optimization effort targets the top. Nearly every real performance problem lives at the bottom, inside the feed and Merchant Center. This note covers the three layers, the failure patterns we see most often, and the order to fix them.

Why Google Shopping Campaign Performance Starts With the Feed

Your product data feed is the most important input in your entire Shopping setup. Google uses it to decide which products match which searches, how your listings look, and how relevant they appear to a given buyer.

The bid comes second. The feed comes first.

Layer	What it covers	Where most teams focus
Layer 1: Product data feed	Titles, descriptions, images, GTINs, pricing, availability, custom labels	Rarely
Layer 2: Campaign architecture	Campaign and ad group structure, product group segmentation, custom labels	Sometimes
Layer 3: Bid strategy and budget	ROAS targets, bidding method, budget allocation	Almost always

If the feed is generic or partially broken, your products show for the wrong queries or not at all. Layer 2 cannot segment products by margin without custom labels set in Layer 1. Layer 3 is fine-tuning output from two layers that may already be broken.

Three Failure Patterns That Hold Shopping Campaigns Back

These are the patterns we encounter most often when we dig into underperforming Shopping accounts. Each sits in a different layer, but they share one trait: the dashboards still look acceptable.

1. The Untouched Feed

The feed was pulled from Shopify or the CMS at launch and never revisited. Product titles use internal naming rather than how buyers search. A product listed as "HRT-CRW-NVY-M" instead of "Men's Navy Cotton Crew Neck Sweater" will never match high-intent Shopping queries.

Google uses the product title as the primary signal for matching Shopping ads to search queries. Generic titles push your ads toward broad, low-converting terms and away from the specific searches that drive revenue.

We see this consistently. A brand running $30K or more per month on Shopping, with feed titles that contain brand codes, abbreviated colors, and no mention of the product type a buyer would actually type. The campaign is running. It is running on bad data.

2. The Performance Max Black Box

The store migrated from Standard Shopping to Performance Max. Reported ROAS improved. Nobody checked what PMax was doing underneath.

Performance Max gravitates toward branded queries because they convert easiest. Without brand exclusions enabled, PMax bids on your own brand terms, captures traffic you would have won organically, and inflates reported ROAS with conversions that were already yours.

PMax also maximizes revenue, not profit. It treats a $10 accessory at 5 percent margin the same as a $200 product at 60 percent margin. Without custom labels in your feed to segment by margin tier, PMax allocates budget wherever it sees the highest conversion probability, which often means your best-margin products are starved of spend.

One detail many brands miss: as of October 2024, PMax no longer has automatic auction priority over Standard Shopping. Ad Rank now determines which campaign wins. A well-structured Standard Shopping campaign can outperform PMax on specific products, but only if the feed data supports the architecture.

3. The Merchant Center Disapproval Leak

This is the quietest failure pattern and one of the most expensive. A meaningful portion of the catalog is disapproved in Merchant Center at any given time, from price mismatches and missing GTINs to images below the minimum resolution.

These products exist in the feed but are invisible to buyers because Google has flagged them.

Google's 2026 Merchant Center specification update raised the minimum image resolution to 500x500 pixels across all categories. Warnings began in April 2026, with full enforcement from January 2027. Brands that have not audited their image assets are accumulating disapprovals they do not know about.

The compounding effect hits hardest during peak trading. Disapproval rates spike during promotions when price changes create mismatches and stock updates create conflicts. This intersects with broader ecommerce technical SEO issues around data freshness and feed-to-site synchronisation.

What to Fix First in Your Shopping Campaign Optimization

Start at the bottom layer and work up. The sequence matters.

Audit your Merchant Center health first: log in to Merchant Center and check product diagnostics. Look at disapproval rates, warnings, and the Needs Attention tab. Every disapproved product has zero visibility. Fix price mismatches, update images to meet the 500x500 minimum, and resolve GTIN issues before touching anything in Google Ads.

Restructure product titles for search intent: the product title is the most impactful attribute in your feed. Use a clear taxonomy: Brand, Product Type, Key Attributes (color, size, material). Run a search terms report in Google Ads and compare what buyers searched against what your feed titles say. The gap directly explains why impression share is lower than expected.

Add custom labels for commercial segmentation: custom labels let you tag products by margin tier, seasonal relevance, performance history, or stock status. Without them, your campaign structure cannot distinguish between high-value and low-value SKUs. Start with margin-based labels; they immediately expose where budget is being misallocated.

Check PMax for branded cannibalization: enable brand exclusions at the campaign level. Compare your branded Search campaign's impression share before and after PMax launched. If branded impression share dropped while PMax ROAS looks strong, cannibalization is happening. Google's brand exclusion feature now covers Search and Shopping surfaces within PMax and catches misspellings automatically.

Build a deliberate campaign architecture: Standard Shopping for transparency on products where query-level data matters, Performance Max for scale on products with proven conversion data. Google recommends at least 30 conversions per month for Smart Bidding to optimize effectively. The architecture follows from the feed segmentation. If your feed is not labeled, your campaigns cannot be structured. We cover how Google Shopping rankings flow from the same feed decisions, which is worth reading alongside this.

Final Thought

Google Shopping campaigns are a data infrastructure problem before they are a campaign management problem.

The feed determines what Google can show. Campaign architecture determines how budget flows across the catalog. Bid strategy fine-tunes the output. Most brands spend their optimization time on the last layer while the first two quietly leak revenue. The stores pulling the strongest Shopping returns have product data that gives Google something clear, specific, and commercially structured to work with. Every month the feed goes unaudited, the gap widens.

Feed and Merchant Center quality are a technical SEO problem before they are a bidding one, which is why we fix the data layer first.