Google Shopping Feed Optimization: What Actually Moves Performance

Most Shopping and PMax audits find the same problem: the campaign isn't underperforming because of bids or budgets. It's underperforming because the feed is mediocre, and mediocre feeds match to mediocre queries, attract mediocre clicks, and teach the algorithm nothing useful.

Agencies rarely go deep on feed work because it's unglamorous and time-consuming. But this is where the leverage is. A cleaner feed doesn't just improve impression share — it changes which queries you show up for, and that changes everything downstream.

This guide is about what actually moves the needle. We'll go attribute by attribute, with real examples.

How Google Uses Your Feed to Match Auctions

Before fixing anything, you need to understand what Google is actually doing with your feed data.

Google Shopping doesn't work like Search. You're not bidding on keywords. Google reads your feed — title, description, category, product type, attributes — and infers which search queries your product is relevant for. The feed is your targeting.

When someone searches "running shoes for flat feet men size 11," Google looks across every Shopping-eligible product and decides: does this product match this query? That decision is based almost entirely on:

1. Title — the highest-weight signal for query matching
2. Description — secondary match signal, lower weight than title but still relevant
3. Google product category — used to understand product context and restrict/enable certain query types
4. Product type — your own taxonomy, used by Google as additional classification signal
5. Attributes — color, size, material, gender, age_group etc. These filter out irrelevant queries and match specific modifier searches

The practical implication: if your title is weak, no amount of bid strategy or campaign structure will compensate. Google can't match your product to the right queries because you haven't told it what your product is.

Title Optimization: The Highest-Leverage Feed Attribute

Your product title is doing the heaviest lifting. Google uses it as the primary text signal for auction matching. Get this wrong and you're invisible for your best queries. Get it right and you'll pick up long-tail traffic you'd never have thought to target.

The structure that works

For most product categories, the proven title structure is:

Brand + Product Type + Key Differentiators + Variant

The order matters. Put the most important keywords earlier. Google weights the front of the title more heavily, and on the results page, titles get truncated on mobile — you want the essential information in the first 70 characters.

Before/after examples

Apparel:

• Before: Women's Blue V-Neck Tee
• After: Lululemon Women's V-Neck T-Shirt — Moisture-Wicking, Slim Fit, Blue, Size S-XL

What changed: brand name added (high-value query signal), product type made explicit ("T-Shirt" not "Tee"), key functional attributes added (moisture-wicking), variant info included.

Consumer Electronics:

• Before: Wireless Headphones Black
• After: Sony WH-1000XM5 Wireless Noise-Cancelling Headphones — Over-Ear, 30Hr Battery, Black

What changed: model number added (critical for high-intent, bottom-funnel searches), key features named explicitly ("noise-cancelling" is a query people actually search), battery life included.

Home goods:

• Before: Standing Desk
• After: Flexispot E7 Electric Standing Desk 60x24 — Height Adjustable, Dual Motor, Black Frame

What changed: brand + model, dimensions (people search these), key specs ("dual motor" differentiates from cheap options), color.

What to avoid in titles

Don't stuff promotional copy. "SALE — Best Standing Desk — FREE SHIPPING" won't get you disapproved, but it wastes characters and dilutes your keyword density with noise words. Promotional language belongs in promotions, not titles.

Don't repeat the same word. "Men's Men Running Shoes for Men" is spam. Google knows it. Save the characters.

Don't use all caps. Legitimately disapproved for many categories, and it looks cheap on the results page.

Don't include your store name unless it's also the brand name. "EcoShop Yoga Mat" adds nothing if EcoShop is your store, not the brand.

Don't leave the default title from your e-commerce platform. Shopify often generates titles like "T-Shirt | Store Name" or pulls whatever's in your product admin. These are almost always suboptimal.

Practical title audit

Pull a Search Terms report from your Shopping campaigns. Look at the high-converting terms. Are those terms actually in your titles? If not, you're getting lucky, and you're leaving volume on the table. Add the missing keywords.

Product Type vs. Google Product Category: Different Jobs, Both Matter

This is one of the most misunderstood areas of feed management. Most operators either ignore product_type, set it to something generic, or confuse it with google_product_category. They're different attributes with different functions.

Google Product Category (GPC)

This is Google's taxonomy. You pick from a fixed list of roughly 6,000 categories (e.g., Apparel & Accessories > Clothing > Activewear). Google uses this to:

• Apply category-level policies (e.g., adult content restrictions, age requirements)
• Understand what type of product it's working with for query matching
• Enable or restrict certain ad features based on product type

You don't have full flexibility here. You pick the closest match from Google's taxonomy. Wrong category = wrong policy rules, potential disapprovals, and weaker relevance signals.

Common mistake: Using a parent category instead of a leaf node. Electronics is far weaker than Electronics > Communications > Telephony > Mobile Phones. The more specific your GPC, the better Google's understanding of your product.

Product Type

This is your taxonomy. It's freeform. You define the hierarchy yourself, and it serves two purposes:

1. Additional classification signal for Google — Google ingests it as an additional text signal. More specific = better.
2. Campaign segmentation for you — This is the big one most people miss.

You can use product_type to build campaign structures. If you set product_type to something like Running > Road Running > Men's > Premium, you can later create Shopping campaigns with product filters based on these values, and bid differently by product type.

Before/after examples

Weak:

• google_product_category: Apparel & Accessories
• product_type: Shoes

Strong:

• google_product_category: Apparel & Accessories > Shoes > Athletic Shoes
• product_type: Running > Road Running Shoes > Men's > Stability

The strong version gives Google better query matching context, enables the right ad policies, and gives you a bidding handle you can pull.

Custom Labels: Your Bidding Segmentation Tool

Custom labels (custom_label_0 through custom_label_4) are five freeform fields you control entirely. Google doesn't read them for auction matching. They exist for one purpose: letting you segment your products for bidding and reporting.

If you're not using custom labels, you're managing Shopping campaigns with one hand tied behind your back.

Strategic ways to use custom labels

Margin tiers. This is the highest-ROI use. If you have high-margin products and low-margin products, you need different ROAS targets for each. Without custom labels, you can't bid them separately. Label your products margin_high, margin_medium, margin_low (using real cut-offs from your cost data), then create separate asset groups or campaigns with different tROAS targets.

Seasonality flags. Label products seasonal_summer or seasonal_winter. When you're heading into a seasonal peak, you can quickly pull up and budget those products without rebuilding your campaigns.

Inventory levels. Label products stock_low when inventory drops below a threshold. You can then suppress or reduce bids on those products automatically, stopping you from spending ad budget on things you can't fulfill.

Price competitiveness. If you track competitor pricing (via a repricing tool or manual monitoring), label products price_competitive vs price_above_market. Bid less aggressively on products where you're priced out.

Bestseller tiers. Pull your top 20% revenue-generating products. Label them bestseller. These warrant more aggressive bids and should be in their own campaign with room to spend.

New products. Label new arrivals new_arrival for the first 30 days. You may want to push spend on these to gather performance data faster, or you may want to protect them from over-spending until you know conversion rate.

The key principle

Custom labels turn your product catalog into a structured, biddable hierarchy. Without them, Shopping campaigns treat a $5 phone case and a $500 keyboard as identical bidding objects. That's a terrible strategy.

Feed Health in Merchant Center: What to Fix First

Merchant Center's diagnostics tab is where your feed problems live. Most operators glance at it, see some red, feel vague anxiety, and close the tab. That's a mistake.

The priority order

Disapprovals first. A disapproved product isn't serving. It's generating zero impressions, zero clicks, zero revenue. Disapprovals are binary — fix them or lose that product entirely.

Common disapproval reasons:

• Mismatched price or availability — Your feed says $49.99 but your landing page says $59.99. Fix it by ensuring your feed refreshes often enough to catch price changes (more on this below).
• Invalid value — A required attribute has a value that doesn't match accepted formats or enum values. Check gender, age_group, condition.
• Prohibited content — You're trying to sell something in a restricted category without the right approvals.
• Policy violations — Review the specific policy cited. Most are fixable.

Account-level suspensions second. These take down your entire account, not just individual products. They're usually triggered by trust signals: mismatched business information, deceptive site practices, or prohibited content patterns. These require careful reading of the policy violation and direct response.

Warnings third. Warnings don't stop your products from serving, but they often mean Google is making do with incomplete data. Common warnings include missing GTINs (where expected), missing size or color for apparel, and low-quality images. Fixing warnings typically improves matching quality.

What to actually fix first (triage)

1. Fix all disapprovals on your top 20% revenue products first. Get your bestsellers live before worrying about the long tail.
2. Add missing GTINs where you have them. GTIN-matched products get better auction eligibility.
3. Fill in missing attributes for apparel and variants. Size, color, gender, age_group are required for apparel — missing these means weaker matching and potential disapprovals.
4. Clean up availability mismatches. These are almost always caused by infrequent feed refreshes.

Images: Technical Specs and What "Better" Actually Means

The image is the first thing a shopper sees. It drives CTR before your title, before your price, before anything else. Yet most operators treat images as an afterthought.

Technical requirements

• Minimum: 100 x 100 pixels (500 x 500 for apparel)
• Recommended: 800 x 800 or larger. Google can zoom in for detail view.
• Maximum file size: 16MB
• Formats: JPEG, WebP, PNG, GIF (non-animated), BMP, TIFF
• No watermarks, promotional overlays, or text on the main image (use additional_image_link for lifestyle shots with text if needed)
• White or light background for most categories — this is what Google prefers and what converts

What "better images" actually means for CTR

Better images aren't just cleaner — they're more scannable at small sizes. When your product appears in a grid of competitors, you have roughly 80x80 pixels of actual attention. Ask yourself: does my product look distinctive and clear at that size?

Specific improvements that move CTR:

• Eliminate busy backgrounds. A lifestyle photo with a kitchen background might look great on your PDP. On a Shopping results page surrounded by competitors, it becomes visual noise. White or light grey backgrounds make products pop.

• Fill the frame. If your product occupies 40% of the image and the rest is white space, you're wasting pixels. Crop tighter. The product should fill 80%+ of the frame.

• Show the right angle. A front-facing flat lay of a shoe converts worse than a 3/4 angle that shows the profile. Test different angles on your highest-volume products.

• Use additional_image_link. You can submit up to 10 additional images. Google can show these in rich product listings. Lifestyle images, detail shots, and size comparison images all add context for high-intent shoppers.

• Check what competitors show. Run your top query in Google Shopping and study the product images that are performing. That's your benchmark, not your brand guidelines.

Feed Refresh Frequency: When Stale Data Kills Performance

Your feed is a snapshot of your catalog at a point in time. Every hour that passes without a refresh is an hour where your live site and your feed can drift apart.

What goes stale and why it matters

Price changes. If you run a sale and your feed hasn't updated, Google may show the pre-sale price in your ad — and then show the sale price on the landing page. This is a policy violation (mismatched price) and a conversion rate killer even when it isn't caught by Merchant Center.

Availability. You sell out of a variant. The feed still says "in stock." You're bidding money on clicks to a product that can't convert. Multiply this by a catalog of thousands of SKUs and it becomes a serious waste.

New products. You launch 50 new products. If your feed refreshes once a day, those products won't be eligible to serve for up to 24 hours. For time-sensitive drops or trend-driven inventory, that's real lost opportunity.

Refresh recommendations

• Daily at minimum. This is the baseline. Any e-commerce operation should be able to get to a once-daily feed refresh.
• 2–4x per day for catalogs with frequent price changes or high SKU turnover.
• Real-time or near-real-time (via Content API) if you're running flash sales, time-limited deals, or managing a large inventory that shifts daily.

Most feed management tools (DataFeedWatch, Feedonomics, GoDataFeed) support scheduled refreshes. If you're on Shopify, the Google & YouTube channel app refreshes the feed periodically, but it's often not frequent enough. Check what schedule your current setup uses — most operators don't know.

5 High-Impact Fixes You Can Make in One Day

If you've read this far and want to prioritize, here's the ordered list:

1. Rewrite the titles on your top 20 products. Pull your Shopping search terms report. Identify your 20 highest-revenue or highest-impression products. Check their titles against the format: Brand + Product Type + Key Differentiators + Variant. Rewrite any that are thin, generic, or missing obvious keywords. This is a 2–3 hour task that compounds over weeks as the algorithm collects better match data.

2. Fix all disapprovals in Merchant Center. Open the Diagnostics tab. Filter by "Item issues." Export the list. Fix every disapproval affecting your top products today. Most disapprovals are a 5-minute fix (update a value, correct a price, fix a policy issue). Don't let revenue-generating products sit disapproved.

3. Add or fix GTINs. For branded products you resell, GTINs are often required and always preferred. Check your Merchant Center diagnostics for missing GTIN warnings. Pull your product list and add ISBNs, UPCs, or MPNs for any product where you have them. This unlocks better auction eligibility and competitive pricing data.

4. Replace the 5 worst-performing product images. Look at your Shopping campaigns by product. Find the 5 products with the highest impressions but lowest CTR. Open their Merchant Center listings. Look at the images. Are they small? Cluttered? Dark? Swap in a clean, white-background, fill-the-frame image and monitor CTR over the next 2 weeks.

5. Set up custom labels by margin tier. Even a rough three-tier segmentation (margin_high, margin_medium, margin_low) gives you an immediate bidding upgrade. Pull your product cost data, calculate margin for each SKU or category, assign labels, and push the updated feed. Then adjust your campaign or asset group tROAS targets accordingly.

Don't Guess — Audit Before You Fix

Every catalog is different. Your title issues might be worse than your image issues, or vice versa. Disapprovals might be concentrated in one product category that accounts for 5% of your catalog and 40% of your potential revenue.

Before you spend time fixing things at random, you need a structured read of what's actually broken.

We built a Google Shopping Feed Audit Checklist that walks you through every attribute category systematically — title structure, image quality, category accuracy, attribute coverage, Merchant Center health, and feed freshness — with scoring criteria so you can prioritize fixes by impact.

Download the free checklist → [link to lead magnet]

If you want us to run the audit for you and come back with a prioritized fix list, that's what we do. [Book a free 20-minute feed review here →]

Track A Guide #3 — Ads Proof