Why the product feed matters more now
A product feed is the structured version of your catalog: one row per product, with the same fields every time. Shopping surfaces on Google read it. AI shopping features that suggest, compare and recommend products depend on product data of this kind too. Where each assistant gets its product data differs and changes, but the principle holds. A system that has to compare a hundred products cannot read a hundred paragraphs. It reads fields.
That makes your data quality a competitive fact. Two stores can sell the same blender at the same price. The one with a clear title, a correct identifier, accurate stock and a precise description is the one a system can confidently show. The one with "Blender 2" and a blank brand field is the one it skips.
This page covers the feed side of Shopify: how data gets from your admin to shopping surfaces, which fields matter first, and how to find errors. It does not cover how assistants choose among stores in general. For that, see AI shopping and product discovery. And for choosing help on the whole of your SEO, our guide to Shopify SEO experts sets out the options.
Your feed is only as good as the product data in your Shopify admin. Fix the data at the source and every surface that reads it gets better at once.
How product data travels from Shopify to shopping surfaces
Understanding the path tells you where to fix things.
Your admin is the source. Product title, description, price, images, variants, inventory and the product type and vendor you enter. Everything downstream is built from it.
Shopify's Google and YouTube channel syncs it. This is a free sales channel you add in the Shopify admin. It connects your store to a Google Merchant Center account and sends your catalog across, updating as products change.
Merchant Center checks and holds it. Google reviews each product against its data rules and marks it approved, limited or disapproved. Approved products can appear in free listings on the Shopping tab, and in ads if you choose to run them.
Page markup is a second, separate source. Your theme outputs product markup on each product page. Google compares it to your feed. When price or availability disagree, products can be disapproved.
That last point is the one stores miss. There are two copies of your product facts, the feed and the page, and they must match. Our page on Google Merchant Center SEO goes through the Merchant Center side in detail.
Which fields matter first
Not every field has equal weight. Work in this order.
Price, availability and condition
These must be correct and must match the product page. A wrong price is the most common reason for disapproval. Out-of-stock items should show as out of stock, not linger as available.
Identifiers
Most products need a brand and a GTIN, which is the barcode number (UPC or EAN) the manufacturer assigns. These let a system know that your product is the same item as one sold elsewhere, which is how comparison works. If you make your own goods and have no GTIN, Google has a way to declare that, and you should use it rather than inventing a number. Never make up a GTIN.
Title
The feed title is what shoppers see in listings. It should lead with what the product is, then the details a buyer filters on: brand, size, color, material, quantity. "Merino Wool Hiking Socks, Men's Size 9 to 12, Charcoal, 3-Pack" tells a system and a person everything. "Cozy Trail Socks" tells neither.
Description
Plain, factual and specific. Materials, dimensions, what is included, who it is for. Skip promotional lines. Feeds work from the content of the description, not its charm.
Images
A clear main image, usually on a plain background, without text overlays or watermarks. Extra angles help.
Category and attributes
The Google product category and details like color, size, gender and age group. These feed the filters shoppers use. Fill them where they apply.
That order matches the quadrant above. The first two are fast and decide whether you are shown at all. The title is slower and decides whether you are picked.
Titles and descriptions across a whole catalog
Changing one title is easy. Changing 300 is where stores stall, so plan the method.
Decide a title pattern per product type. For socks, it might be product type, then brand, then size and color, then pack size. Write the pattern down. A consistent pattern lets you edit fast and keeps the catalog coherent.
Edit in bulk where Shopify allows it. The product list can be exported and imported as a spreadsheet. You can rewrite titles in a sheet using formulas to assemble the pattern, then import the changes. Always export a backup first and test with five products.
Mind the two titles. Your product's page title for Google Search, set in the search listing section, and the product title that flows into the feed are not always the same field. The first is for ranking in search results, the second is what shopping surfaces read. They can differ, and each should be written for its own job. Our page on product page SEO covers the page side.
Start with the top 50. Rank products by revenue or by impressions in Merchant Center and fix those first. Do not stop everything to polish the long tail.
Keep claims honest and checkable. A title that says "waterproof" must be true of the product, and a mismatch between feed and page text can trigger a policy flag.
Finding and fixing feed errors
Merchant Center has a diagnostics area that lists every problem it found, grouped by type. Open it before you change anything.
Disapproved items. These will not show at all. Read the reason. Typical ones: price on the page does not match the feed, a missing required attribute, or a landing page that does not work.
Warnings. The product shows but may be limited. A missing GTIN or an image Google thinks has an overlay is common here.
Suggestions. Optional fields that would help matching or filtering.
Fix by cause, not by product. If 80 products are disapproved for the same reason, there is usually a single cause: a theme edit that broke markup, an app that rewrites prices, a currency setting. Find that and the whole group clears.
Then use the free Rich Results Test on one affected product page to confirm the on-page markup says what the feed says. Our page on free Shopify SEO tools shows how to use it.
What AI shopping does with product data
Be careful here, because this area changes quickly and anyone claiming certainty is guessing. What can be said sturdily:
Assistants that recommend products need structured facts. Price, availability, specifications and images arrive from feeds, from page markup and from retailer pages. Better, cleaner data means more chance of being included accurately.
Descriptions get compared. When someone asks for "the best backpack for a 15 inch laptop under a certain price", a system looks for products whose data states capacity, dimensions and price. A product with those stated is eligible. One without them is guesswork.
Reviews and outside sources count too. Product data says what an item is. Independent opinions say whether it is good. Assistants lean on both, which is why our page on reviews and AI recommendations belongs next to this one.
You cannot buy your way in through the feed alone. A clean feed makes you eligible. It does not make you chosen.
Where each assistant sources its shopping data is not fully public and can change. Treat any single-source claim, such as "this one feed unlocks ChatGPT shopping", with caution. The safe strategy is the one in this page: accurate data at the source, matching markup on the page, and a good reputation off the page.
Keeping the feed clean over time
Feeds decay. You add products, change prices, discontinue lines and install apps. A feed that was perfect in March has errors by September.
Set a monthly check. Open Merchant Center diagnostics. Note the count of disapproved and warning items. If it rose, find the cause.
Add data at product creation. When you add a product, fill in brand, product type and identifiers right then. It is much cheaper than backfilling later.
Be careful with apps that touch prices. Discount and currency apps can change the displayed price without changing the feed, which creates a mismatch.
Write a data standard. A short document listing how titles are built, what goes in the description and which fields are mandatory. It keeps anyone adding products consistent. If you outsource any of this, our page on Shopify SEO shows where it sits among the other work.
The benefit compounds. Every surface that reads your catalog gets slightly better, and your own site search and filters do too.
When to do it yourself and when to get help
Under about 100 products, the feed is a weekend of work and then a monthly check. You can do it. Start with the diagnostics list and the top-left of the quadrant.
Above a few hundred products, or with many variants, the volume becomes the problem. The decisions are simple. The repetition is not. That is when a spreadsheet method, a developer or a service earns its cost.
Get help sooner if you sell regulated products, if your catalog has no barcodes because you make everything yourself, or if disapprovals keep coming back after you fix them. Those usually trace to something in the theme or an app, and finding it takes some patience.
A worked example: one product, before and after
Abstract advice is easy to nod at and hard to apply, so here is one invented product taken through the process.
Before. A canvas backpack. Feed title: "Trail Pack". Description: "Our best seller! Loved by everyone." No brand filled in. No GTIN, because the store makes it. Price in the feed is $89 while the product page shows $79 after a sale. Main image has a "SALE" badge in the corner.
What a system sees. An unnamed item with no category clue, a price that disagrees with the page, an image with promotional text, and a description that states nothing checkable. It is likely to be limited or disapproved, and even if shown, it is a poor candidate to match to a shopper's question.
After. Title: "Waxed Canvas Backpack, 22 Liter, Fits 15 Inch Laptop, Olive". Description: materials, volume, dimensions, weight, number of pockets, laptop sleeve size, who it suits, care instructions. Brand set to the store name. Identifier declared as not available because the store makes the product. Price in the feed matches the page, and the sale is handled through a proper sale price field. Main image clean, on a plain background, with the badge moved to a second image.
What changed for a system. It now knows what the thing is, how big, what it fits, and what it costs, and the facts agree in both places. A question like "waxed canvas backpack that fits a 15 inch laptop" has a product that plainly matches.
Nothing in the example needed a developer. It needed a person who knew the product and an hour.
How RunOctopus handles this
The feed is data work. The content that sits around it, the guides and pages that explain why your product is the right one, is the writing work.
RunOctopus is a done-for-you service. Its AI, Ollie, researches your business, then writes and publishes pages directly on your own website. A team plans, writes, reviews and publishes alongside Ollie. It works with Shopify, WordPress, WooCommerce and Wix.
There is nothing to install, nothing to approve, nothing to upload and nothing to schedule. You do not approve each page. Pages go live on their own. In the first week you get 8 long-form articles, 6 pages and a custom interactive tool built for your niche. After that it is 175+ pieces of content a month, written, published and linked. You also get reports on where ChatGPT, Perplexity, Claude and Gemini name your business.
There is a 90-day promise: more visibility on Google and in AI answers within 90 days, measured against the 90 days before you joined, or every dollar back. For scale, an agency doing this work charges about $3,209 a month (2026 average, 439 providers) and publishes 4 to 12 pieces.
You can see what Ollie would do for your store before you decide anything. Paste your address into the free site read. Ollie reads the site and shows the plan. No account, no call.