Bulk edit Shopify SEO by CSV without breaking your catalogue

How Shopify’s product CSV import handles updates

Shopify’s product import does two jobs with the same file: it creates new products and it updates existing ones. What decides between the two is the handle, the unique identifier that also forms the product’s URL.

When you import, Shopify offers an option called Overwrite products with matching handles. With it ticked, any row whose handle matches an existing product overwrites the values in the matching columns. Without it, products that match an existing handle are ignored during the import. So for an SEO update, the option has to be on, and that is where the risks begin.

Three rules from Shopify’s documentation govern what an overwrite does:

What is in your file	What happens to the live product
Column present, cell filled	Value replaced by the cell
Column present, cell blank	Value overwritten as blank
Column absent from the file	Value left unchanged

There is a fourth, less obvious rule: if you include a column that depends on other columns you left out, existing data can be deleted. Shopify gives the example of variant columns such as SKU without their option columns.

Two more constraints to know before you start: the file can’t exceed 15 MB, and a product import can’t be cancelled once it begins.

The columns that matter for an SEO update

For updating SEO fields, you need very few columns:

• URL handle: the match key (older exports call it Handle). Never edit it in this file.
• Title: required when updating. Keep the exported value unchanged.
• SEO title: the page title shown in search results. Shopify’s CSV documentation allows up to 70 characters.
• SEO description: the meta description. Up to 320 characters.

Everything else (description, vendor, tags, variant, price and image columns) can be deleted from the file. Since absent columns are left untouched, a four-column file is the safest possible SEO import: it physically can’t blank a price or detach an image.

A note on empty SEO cells. Blanking the SEO title does not leave the product without a title in search results: Shopify falls back to the product title and description. But the result is a snippet built from copy written for the product page, which is rarely what you wanted. Treat every blank SEO cell as a deliberate decision, not a leftover.

A note on handles. Because the handle is the match key, you can’t rename a product’s URL through the import by editing the URL handle column: the row stops matching the original product. Change handles in the admin, where you can create a redirect from the old URL.

Excel, encoding and other ways a file goes wrong

Most CSV disasters happen in the spreadsheet, not in Shopify.

• Encoding. Shopify requires UTF-8 to prevent unexpected characters. In Excel, save as “CSV UTF-8 (Comma delimited)”, not plain “CSV”, or accents in “Crème” and “Größe” turn into garbage across the whole catalogue.
• Separators. Shopify’s help center asks you to check Excel’s export settings so values are separated by commas. Some regional Excel settings default to semicolons.
• Sorting. Shopify warns that sorting the file in a spreadsheet editor can break the link between products and their images. If you sort to work more comfortably, don’t import the sorted file with image columns in it.
• Formulas. A =CONCAT(...) used to build meta descriptions must be converted to values before export.
• Header names. Headers must match Shopify’s template exactly, so don’t rename or translate them.

A safe procedure for 500 products

Say you run a store with 500 products and want new meta descriptions on all of them. Here is the order of operations that keeps the blast radius small.

1. Export all products to CSV as a backup, and keep it untouched
2. Duplicate the export as your working file
3. Keep only URL handle, Title, SEO title, SEO description
4. Write or paste the new SEO copy
5. Check lengths and look for blank SEO cells
6. Save as CSV UTF-8 with comma separators
7. Import 5 rows first, with overwrite ticked
8. Open those 5 products and compare with the backup
9. Import the remaining rows
10. Re-export and diff against the backup file

Step 10 is the one people skip. A plain spreadsheet diff (or a diff in the terminal on the two sorted exports) shows every field that changed. If anything outside the SEO columns moved, you know right away, and your backup is the file to restore from.

When the bulk editor is the better tool

For a few dozen products, Shopify’s bulk editor can be simpler than a CSV round trip. Select products in the Products list, click Bulk edit, and use Columns to add the properties you want to edit; metafields can also be edited there. Check that the fields you need are offered as columns in your admin. The bulk editor works inside the admin, so there is no encoding or separator to get wrong, but it gives you no before and after comparison beyond what you remember.

Above a hundred rows, or when the new copy comes from an agency, a writer or an AI tool in a spreadsheet, the CSV route is usually faster. It just needs the checks above.

Sending the CSV to review instead of straight to the storefront

The procedure above has a weak point: step 9 publishes 495 changes at once, and review happens afterwards, if at all. For SEO copy written by someone else (a freelancer, an agency, an AI tool), you would rather read the changes before they go live.

Vetta’s CSV import is built for that case. You map the spreadsheet’s columns to fields once (SEO title, meta description, tags, metafields, and translated fields with one column per language), and the mapping is saved for the next import. Before anything runs, the import shows an estimate of how many versions it will create and what it will cost. Each row then becomes a proposed version, not a live change: reviewers see a field-level diff of live against proposed, word by word, and approve, request changes or reject each one. Nothing reaches the storefront until it is approved.

Because every version is kept, including edits made directly in the Shopify admin, rolling back a bad batch means restoring the previous versions, not hunting for last month’s export. Credits are charged per approved and published version, which the pricing page explains.

Whichever route you choose, the rule is the same: keep the file small, keep a backup, test on a few rows and look at the diff before you trust the import.