Most Shopify stores have the same SEO gap: a few hundred products where the SEO title and meta description were never written. Shopify fills in defaults, so nothing looks broken. The page title is the product name, and the description is whatever the first sentence of your product copy happened to be.
Writing them properly is the definition of a job nobody wants. It's repetitive, it follows rules, and there are hundreds of them. That makes it a very good job for AI. Here's a workflow that does a whole catalog in an afternoon, and keeps you in control of what gets published.
Why the CSV route
You could ask an AI assistant to update products one by one through a store connection. For this job I'd use a spreadsheet instead, for three reasons:
- You can see all of it before any of it goes live. Every title, in a column, sortable by length.
- It scales. Shopify's official AI integrations change items one at a time, and large batches can stop partway. An import handles thousands.
- You keep a backup. The original export is your undo.
A Shopify product export includes SEO Title and SEO Description columns. Fill those, import, done.
Step 1: Export
In your Shopify admin, go to Products and export to CSV. Export everything, or filter to one collection to start small. Starting small is worth it: do 20 products end to end before you do 2,000.
Save an untouched copy of the file somewhere safe.
Step 2: Give AI the rules
Upload the CSV to Claude or ChatGPT with a prompt that spells out what good looks like. Vague instructions produce 300 titles that all start with "Discover."
For each product in this Shopify export, write an SEO Title and an SEO Description.
SEO Title: 50 to 60 characters. Start with the product name. Add the one detail a shopper would search for, such as material, size, or use. End with our brand name, [Brand], if it fits.
SEO Description: 140 to 160 characters. One or two plain sentences. Say what the product is and the main reason to buy it. Use facts from the Title, Body (HTML), Type, and Vendor columns only. Do not invent features, materials, or claims.
No exclamation marks. No "discover," "must-have," "perfect for," or "look no further." Don't repeat the title in the description.
Only fill rows where the Title column has a value. Leave every other column exactly as it is. Add two columns at the end with the character count of each.
A few of those lines are doing a lot of work:
- "Facts from these columns only" stops it inventing a feature to make a sentence more interesting.
- The banned phrases remove the tells that make AI copy read like AI copy.
- "Only fill rows where the Title column has a value" matters because a Shopify export has several rows per product, one for each variant and image. The SEO fields belong on the first row.
- The character counts make your review much faster.
Step 3: Review it properly
Open the result in a spreadsheet. Don't read every row top to bottom. Sort and scan.
Sort by title length. Fix anything well over 60 characters. Look hard at anything under 30, which usually means the detail got dropped.
Sort by description length. Same again around 160.
Look for duplicates. Two products with the same SEO title compete with each other. Variants of a similar product are the usual offenders: the titles need the thing that distinguishes them.
Check facts on a sample. Pick 15 rows across different product types and compare each claim to the product. If the description says "organic cotton," the product had better be organic cotton.
Read ten aloud. If they sound like the same sentence with the noun swapped, send it back: "These are too similar. Vary the structure, and lead with what's different about each product."
Check your best sellers by hand. Your top 20 products earn most of your search traffic. Write or edit those yourself.
Step 4: Import
Back in the admin, import the edited CSV and choose to overwrite existing products with matching handles. Shopify shows a preview first. Read it. Check a couple of rows look right before you confirm.
Then open three products in the admin and look at the search engine listing section to confirm the new text is there.
What a good one looks like
Take a product called "Harlow Tote" with a description that begins "Our bestselling bag is back."
Default (what Shopify uses if you write nothing):
- Title: Harlow Tote
- Description: Our bestselling bag is back. We've listened to your feedback and...
Written:
- Title: Harlow Tote, Full-Grain Leather Work Bag | Brand
- Description: A structured leather tote that fits a 15-inch laptop, with an inside zip pocket and a detachable strap. Made in small batches.
The second one tells a searcher what it is, what it's made of, and whether their laptop fits. The first tells them there's a bag.
Collections and pages too
Collections and pages have SEO fields as well, and collection pages often matter more for search than individual products. There are fewer of them, so do these by hand with AI as a drafting partner:
Here's my collection name, its description, and the ten products in it. Draft three options for an SEO title and meta description.
Pick the best and edit it. More on what makes these pages rank in making Shopify collection pages SEO-friendly.
What this fixes, and what it doesn't
Be realistic about the size of the win. Better titles and descriptions improve how your listing reads in search results, which can lift clicks. They won't rescue a page with thin content or a product nobody searches for. And Google rewrites snippets when it thinks it can match the query better, so your description won't always be the one shown.
It's still worth doing. It's cheap, it's a one-time job per product, and a clear title is the first thing both a person and a machine learn about the page.
For AI search specifically, titles and descriptions are a supporting act. Assistants like ChatGPT and Gemini rely more heavily on structured product data: the product's name, price, availability, and attributes, published in a form they can parse. We cover that in product descriptions that rank and that AI can parse and the Shopify technical SEO checklist.
If you're already cleaning up product copy, it's a good moment to check the structured side as well. The free AI readiness checker shows what AI assistants can read from your store in about a minute.
The short version
- Export products to CSV and keep a backup copy.
- Give AI the length limits, the structure, the facts-only rule, and your banned phrases.
- Sort by length, hunt for duplicates, fact-check a sample, and hand-edit your top sellers.
- Import with the preview, then spot-check in the admin.
The same export-edit-import pattern works for other fields. See filling missing product data with AI and bulk editing products with AI.


Comments
Every comment here comes from a verified email. Write yours, confirm from your inbox, and it's live.
Loading comments…