the guide
Shopify product data, explained for a store of a few hundred products.
Google, the Shop app, AI assistants, your store's filters and every channel feed decide whether to show a product by reading its fields. Not your theme, not your ads, the fields. This guide is every one of them: what it is, who reads it, what breaks when it is empty or spelled three ways, and the order to fix it in, without a team or a spreadsheet.
who reads it
You write it once. Six readers decide on it.
The same fields feed every place a shopper can find you. None of them read your mind and none of them read your theme. They read the fields, and an empty field matches nothing.
Google search
Reads the SEO title and meta description for the two lines in results, and the title, description and details for what to rank you for. Left blank, Shopify shows the product title and the first lines of the description, which is why so many stores' results look identical.
Google Shopping and Merchant Center
Reads the category, the vendor as the brand, identifiers, images, and for apparel the size, color, gender and age group. Google's own list of product data violations names each of these: a missing brand, a title in capitals, an invalid category, a generic image, a missing apparel attribute. Each one can get a product shown less, or disapproved.
Your store's filters
Storefront filters are built from variant options, the category's detail fields, and metafields that have a definition. A product with the field blank never appears in that filter's results. A value spelled two ways appears as two choices, each holding part of your catalog.
The Shop app and AI assistants
Shopify Catalog is the feed. Shopify says it is read by ChatGPT, Copilot and the Shop app, that each agent "decides which products to show and how to rank them, weighing the buyer's request, product data quality, availability, price, shipping, trust signals, and other factors," and that "the clearer, more specific, and more current it is, the easier it is for an agent to match and recommend you." To be in it at all a product needs a title, at least one image, a price above zero and a published page.
Screen readers
Read the alt text in place of the image. Blank alt text is a photo that is not there for that shopper, and it is one of the things Google Images uses to know what a photo shows.
every field
The fields, who reads each one, and what breaks.
The six that hide products most often come first, as cards. The full list is the table under them.
Category
What it is. The Shopify category, from Shopify's standard taxonomy.
Why it matters. It decides which detail fields the product gets, and Google uses the category to decide which details it requires.
Fix it →Product details
What it is. The category's own fields: color, size, material, and so on.
Why it matters. Store filters and Merchant Center read these. Empty means filtered out or, for apparel, disapproved.
Fix it →Description
What it is. The text that describes the product.
Why it matters. Shoppers decide on it; Google matches searches against it; AI assistants quote it.
Fix it →Image alt text
What it is. A short line on each image saying what it shows.
Why it matters. Screen readers read it aloud; Google Images and AI use it to know what the photo is.
Fix it →Search title and summary
What it is. The SEO title and meta description, the two lines Google shows.
Why it matters. Left blank, Shopify shows a default and Google shows the fallback. Written, they win the click.
Fix it →Translations
What it is. Every field above, in each language you sell in.
Why it matters. Shopify falls back to your main language wherever a translation is missing, so shoppers see a mix.
Fix it →| Field | Who reads it | What breaks when it is empty or inconsistent | Where it gets fixed |
|---|---|---|---|
| Title | Everything | Weak or all-caps titles rank low and can be demoted by Merchant Center | Title check; product field edit |
| Description | Shoppers, Google, AI assistants | Nothing to read, nothing to match, nothing to quote | Missing description fill |
| Category | Google Shopping, Shop app, AI agents, filters | Not in the category, wrong detail fields offered | Missing category fill |
| Category details | Filters, Merchant Center, AI agents | Filtered out; clothing can be disapproved without color, gender and age group, and size for clothing and shoes | Details fill |
| Vendor | Merchant Center (as brand), filters, feeds | Missing brand is a Merchant Center error; three spellings are three brands | Vendor check |
| Product type | Admin, filters, some feeds | Type list turns into a mess of near-duplicates | Product field edit |
| Tags | Filters on old themes, automated collections, Flow | A spelling difference is a different tag | Tag check |
| Options and variants | Filters, Merchant Center | Size and size become two filters; sizes in three formats | Admin |
| Images and alt text | Screen readers, Google Images, AI assistants | Blank alt text: nothing said aloud, and Google has less to go on | Alt text fill |
| SEO title and meta description | Google search results | The fallback lines, same shape as every other store | Default SEO fill |
| Metafields | Filters (with a definition), feeds, AI agents | Cotton, cotton and 100% cotton become three values | Metafield edit |
| Identifiers (GTIN, MPN) | Merchant Center | Missing or wrong identifiers block approval | Admin; not something Fillwell writes |
| Translations | Shoppers in other languages, Google per locale | A mix of two languages on one page | Translation fill |
"Where it gets fixed" names the Fillwell fill, check or edit when there is one, and "Admin" when there is not. Identifiers, prices, weights, images and variant option values are yours.
what it costs
Empty fields lose the sale. Wrong fields cost the return.
Three sources, two of them consumer surveys from 2025: one of 8,500 shoppers worldwide and one of 1,900 in the US and UK.
abandoned a purchase because the product information was not enough.
have returned a product because it did not match the online listing.
think worse of a brand when its product information is incomplete or wrong. Up from 62% two years earlier.
will spend more on a product with detailed information.
Sources: Syndigo, 2025 State of Product Experience (44%, 75%); Salsify, 2025 Consumer Research (71%); GS1 US, via Shopify's product data enrichment guide (62%).
the audit
Twelve questions. Count the no's.
An audit is not a score. It is a count per field: how many products fail each question. Do it by hand on a catalog under a hundred products, filtering the admin's product list by empty fields. Above that, the scan in the app does the counting.
The free scan reads up to 250 products on your public store and answers the questions on descriptions, alt text, search fields and spellings. The scan in the app reads every product and the rest, including categories and category details.
- Does every product have a category from Shopify's taxonomy, and is it the right one?
- Are the category's detail fields filled: color, size, material, and the rest?
- Does every product have a real description, a few sentences long, in your own words?
- Does every image have alt text that says what the photo shows?
- Did you write the search title and summary, or is Shopify showing the default?
- Is the vendor spelled one way? Is each tag spelled one way? Is each option named one way?
- Do product types repeat with different spellings?
- Are your metafield values consistent, and do filterable ones have a definition?
- For apparel: size, color, gender and age group on every product?
- Do products that need one have a GTIN, or are they marked as custom?
- If you sell in more than one language, is every field translated?
- Is there a written naming rule your team follows?
the fix
Fix in this order. The order matters.
1. Categories
First, because the category decides which detail fields a product has at all. Fill details before categories and you fill the wrong ones. Pick the most specific category that is true; a general one loses the details Google and filters read.
2. Details
Color, size, material and the rest, per product, from what the product already says. This is the one the public scan cannot see, because the fields are not on the storefront. Apparel stores: color, gender and age group are Merchant Center requirements, and so is size for clothing and shoes.
3. Titles and search fields
A title a buyer would type: what it is, then what makes it that one. No sale words, no capitals, no model number first. Then the SEO title and meta description, written, not defaulted. Do these before descriptions because they are short and they are what Google shows.
4. Descriptions
One product at a time, in your words: what it is, what it is made of, who it is for, how it is used or cared for. A pasted supplier description is the same text every other store selling the product has.
5. Alt text
Per image, saying what that image shows. The front view, the size chart and the fabric close-up are different pictures and get different lines. Leave alone anything already written.
6. Translations
Last, because everything above changes the source. Then every field, in every language you sell in, including alt text and the SEO fields.
Any time: consistency
Vendors, tags, product types and option names, one spelling each. Merge the rest. Update any automated collection or Flow rule that named the old spelling before you merge, so it does not empty out. Then add a rule: tags picked from the existing list, never typed; metafields with preset choices turned on in the definition, so Shopify refuses a new spelling at the source.
The scan finds the gaps
Free, every product, on every plan. A count per field, not a hunt.
AI fills, or you set
From the product's own text. Alt text from the image. Not sure? It skips.
You approve
A page at a time, or in bulk with review after. Nothing saves until you say yes. Undo anything.
Fillwell fills the six gaps, checks vendors, tags and titles, and bulk-edits types and metafields on a paid plan. It does not write identifiers, prices, weights or images. See what it fixes, what it checks and pricing.
good to know
Common questions
What counts as product data?
Every field on a product in Shopify: title, description, category, the category's detail fields, vendor, type, tags, options and variants, images and their alt text, the SEO title and meta description, metafields, identifiers, price and inventory, and each field's translations. Google, the Shop app, AI agents, filters and feeds read those fields and nothing else.
Which fields matter most?
Category first, because it decides which detail fields the product gets and it is the first thing Google Shopping, the Shop app and AI agents read. Then the details, the description and the title, which are the words everything matches against. Alt text and SEO fields after that. Vendor and tag consistency matters more than it looks, because feeds and filters treat each spelling as a different thing.
Does Shopify fill any of this for me?
Some. Shopify suggests a category, Shopify Magic drafts a description one product at a time, and Sidekick can draft several. Shopify does not tell you which products are missing what, and it does not check whether your vendors, tags and titles are consistent. That is the finding, and it is the slow part.
How do AI shopping assistants get my products?
Through Shopify Catalog, which Shopify says feeds ChatGPT, Copilot and the Shop app. Each agent decides what to show and how to rank it, and Shopify names product data quality as one of the inputs. The agent reads the same fields as everyone else: title, description, category, details, images, price and availability.
What does bad product data cost?
Syndigo's 2025 survey of 8,500 consumers: 44% abandoned a purchase over insufficient product information, 21% returned a product because the content set the wrong expectation, and 75% think worse of a brand with incomplete information. Salsify's 2025 survey: 71% returned a product because it did not match the listing. Those are returns and lost sales, on top of the products that never showed up at all.
Where do I start on a catalog of a few hundred products?
Run an audit first, by hand with the checklist above or with the free scan, so you know the counts. Then fix in this order: categories, details, titles and search fields, descriptions, alt text, translations. Categories come first because they change which detail fields exist. Consistency checks on vendors and tags can run any time.
What does Fillwell do and not do here?
It finds every product with a missing category, blank details, no description, blank alt text, a default SEO field or a missing translation, fills them from the product's own text, and alt text from the image, and holds each value for your approval with undo. It finds and merges vendor and tag spellings, and flags titles Google asks you to avoid. It does not write identifiers, prices, weights or images, and it does not edit variant option values.
How often should I check?
Every time you add products, and after any import. The scan and the checks are free, so the honest answer is: run them after each batch of new products, and fix what they find while it is a page, not a backlog.
Primary sources: Shopify on Shopify Catalog and its requirements; Google Merchant Center on product data quality violations; Shopify's standard product taxonomy; storefront filters.
you're in control
Approve each change, or bulk fill and undo anything.
AI can do this work. It shouldn't do it alone. Fillwell is built that way on purpose.
- You guide it. Tell it what to do in plain words.
- It skips when unsure. It never guesses into your live store.
- You can undo anything. One row, one page, or all of it, for 30 days.
Your store. Your decision.
Get the material from the @description.
explore fillwell
See everything Fillwell can do.
One free scan finds the gaps. Then Fillwell fills them, and you approve or undo. Start anywhere.
See how clean your product data is.
Scan your store free. See the count for each field. Fix in order.
Free scan · public store only · no install needed.