Google Shopping Feed Optimization: Titles to Custom Labels (2026)

Shopping ads have no keywords. Google decides which of your products to show for a search almost entirely from the data in your product feed. That is why two stores with the same budget can get very different results: one shows up for "women's leather ankle boots", the other barely gets impressions. The difference is rarely the bid. It is usually the title, the category and the fields that were left empty. This guide covers the whole feed with rules you can apply today and worked examples. Menu names are accurate as of September 2026.

What is Google Shopping feed optimization and why does it matter?

Google Shopping feed optimization is the work of shaping the product data you submit to Merchant Center so it fits both Google's matching logic and the words shoppers actually type. Because Shopping and Performance Max campaigns use product data instead of keywords as their targeting signal, feed quality translates directly into impressions and clicks.

In practice the feed controls three things: which searches your product is eligible for, what the ad says, and how you can group and bid on products inside a campaign. The first two depend on the title, description, GTIN and category. The third depends on product_type and custom labels. If you run Performance Max, the same feed is its fuel too; our feed-first Performance Max setup guide covers the campaign side.

How do you optimize Google Shopping product titles?

A strong Shopping title combines the brand, the product type and the attributes that set the item apart (color, size, material, capacity), with the most important words in the first 70 characters. Merchant Center accepts 1–150 characters, but depending on the screen, most shoppers only see the first 70 or fewer.

The most common mistake is sending the "creative" product name from your storefront straight to the feed. Nobody searches for "The Luna Dress". "Reformation Women's Satin Midi Dress Black Size 6" carries the words people search for and the details that separate it from similar items. These patterns work well by category:

Let search behavior decide the order. If your brand is well known, keep it first. If you sell your own label and nobody searches for it yet, lead with the product type and move the brand to the end. Your Shopping search terms report is the best place to see the exact phrasing buyers use.

Each variant needs its own title. If the small and medium versions of the same shirt share one title, Google cannot tell which one fits which search. The color attribute should use exactly the name on the landing page: if the page says "Midnight Blue", send "Midnight Blue" in the color field, not "Navy". In the title, the plain color word people actually search for (navy, black) usually catches more queries.

If you generate titles with AI tools, Google asks you to submit them in the structured_title attribute with digital_source_type set to trained_algorithmic_media rather than in the regular title field. Review AI output before it goes live; it tends to add adjectives that no one searches for.

Which feed attributes matter most besides the title?

After the title, the fields that do the most work are GTIN, google_product_category, product_type, description and the variant attributes (color, size, gender, age_group). They improve matching and make it easier to split campaigns.

• GTIN: If the manufacturer assigned a barcode (UPC in the US, EAN in the UK and Europe), send it. Google reports that retailers who added correct GTINs saw clicks increase by 20% on average. Never invent GTINs for your own unbranded or custom products.
• google_product_category: Go at least 2–3 levels deep, from broad to specific. Not "Apparel", but "Apparel & Accessories > Clothing > Outerwear > Coats & Jackets".
• product_type: Your own category tree, separated by " > ". You can submit several values, but only the first is used for bidding and reporting in Shopping campaigns, so put your most useful grouping first.
• Description: Say what the product is in the first sentence, then give concrete facts: dimensions, material, compatibility, what's in the box. It should match the landing page; Google warns that synonyms or search terms missing from the page can confuse shoppers.
• Variant attributes: Fill in color, size, gender and age_group. For apparel sold in many countries, several of these are required anyway.
• Price and availability: Must match your site at all times. Google states plainly that mismatches cause disapprovals; our guide to disapproved products covers the fixes.

What are the Google Shopping image requirements in 2026?

Your main image should show the product clearly, with no watermark and no text overlay. Google has announced a minimum of 500 x 500 pixels for all products starting January 31, 2027, and recommends around 1500 x 1500 pixels or larger where possible.

The list of banned elements is clear: watermarks, calls to action like "Buy now", price or promotional text such as "free shipping", borders and overlays, barcodes, logos that are not part of the product, and placeholder images. Files can be up to 16 MB and 64 megapixels. Google recommends a solid white or transparent background for the main image (avoid transparent backgrounds for light-colored products, which can render on black). Send lifestyle shots as extra images through additional_image_link. If your product photos carry a sale banner, keep a clean version for the feed and show the banner only on your site.

How do you set up feed rules in Merchant Center?

In the current Merchant Center, feed rules are called attribute rules. You open a product source under Data sources and add rules in its Attribute rules tab, which lets you transform data without editing your XML feed or your ecommerce platform.

You need the "Advanced data source management" add-on turned on (Settings > Add-ons). Rules run as a cascade: the output of the first rule becomes the input of the second. A typical flow looks like this:

1. Pick the source Go to Data sources, open your primary source under Product sources and switch to the Attribute rules tab.
2. Choose the attribute Click "Add attribute rule" and select the field to change, for example title.
3. Define the value Use "Set to" to combine fields, for example title = brand + " " + title.
4. Add conditions and edits Limit the rule with AND/OR conditions, then use Prepend, Append or Find & Replace to change the text.
5. Test and apply Save as draft, use Test rules to check a few products, then click Apply changes.

The most useful rules in practice: prepend the brand to the title, append color and size to the end, strip phrases like "Free Shipping" from descriptions with Find & Replace, and assign a custom label based on price with Calculate or conditions. A rule applied to the wrong products spreads across the whole feed, so keep conditions narrow and test on a small group first.

How should you use custom labels in Google Shopping?

Custom labels are five internal fields (custom_label_0 to custom_label_4) that group products by your own business rules and are never shown to shoppers. Their most valuable use is separating products by margin, season and sales velocity so each group gets its own budget or target ROAS.

Each value can be 1–100 characters, and each label field supports up to 1,000 unique values across the account. You decide what each label means; the key is to dedicate one field to one purpose and keep the number of values small. A layout that works for most stores:

Worked example: targets by margin. Break-even ROAS is roughly 1 divided by your gross margin. A product group with a 40% margin breaks even at a ROAS of 2.5 (250%); a group with a 20% margin needs 5.0 (500%). Say you sell a $100 item at 40% margin: you keep $40, so spending more than $40 in ads per sale loses money. At 20% margin the ceiling is $20. Put both groups in one campaign with a single target and Google will accept losses on the thin-margin products while being too cautious with the profitable ones. Try your own numbers with the break-even ROAS calculator, and see what ROAS is for the underlying logic.

Primary feed or supplemental feed: which should you use?

The products themselves (id, title, price, availability, image) should come from your primary source: an XML or CSV feed, or a platform integration like Shopify's Google & YouTube app. Extra data such as custom labels or rewritten titles is safer to add through a supplemental data source, matched by product id, so the primary feed stays untouched.

If you have more than one source for price or stock, the "Take latest" operation in attribute rules uses the most recently updated value. That helps when a daily XML feed runs alongside an hourly inventory file.

How do you measure the impact of feed changes?

Compare the product group you changed against an equal-length period before the change, looking at impressions, click-through rate and conversion value. Title changes usually show up first in impressions and CTR; custom label changes show up in profitability.

Work in batches rather than rewriting everything at once. Rewrite the titles of your 50 highest-spend products, wait two to three weeks, then compare those same products against the previous period. If you change your bid strategy at the same time, you won't know which change caused what. For the account-level side, the Google Ads optimization guide is a good companion.

What should you do this week?

• List disapproved and warning products in Merchant Center and fix price and availability mismatches first.
• Rewrite the titles of your top 50 products by spend using the category formula; check what fits in the first 70 characters.
• Find products that have a manufacturer GTIN but don't send it, and add it.
• Bring google_product_category and product_type to at least 2–3 levels deep.
• Reserve custom_label_0 for margin and plan separate ROAS targets per margin group.
• Replace images that have watermarks, sale banners or borders, and make sure every image already meets the 500 x 500 minimum coming in 2027.