How to translate a product catalog CSV for e-commerce

The Bottom Line: Translating a product catalog is the single highest-ROI localization task for any e-commerce business: the most efficient workflow is to isolate your translatable fields from your structural metadata and use a glossary-aware engine to ensure your brand’s voice remains consistent across thousands of SKUs.

Using the best website translation software is the only way to scale this safely.

That fear is justified. A product CSV is not a simple document. It is a structured data file where every column has a specific purpose. If your translation process accidentally modifies a SKU, changes a product handle, or shifts a price column, your import will either fail completely or, worse, silently corrupt your store data.

The good news: with the right approach, you can translate thousands of products in minutes without touching a single structural field. This guide covers the entire process, from understanding your catalog’s CSV structure to choosing the right translation mode and handling the tricky edge cases that trip up most teams.

Why product catalog translation matters more than you think

Let me share a number that surprises most e-commerce operators: localized product pages convert 2 to 3 times better than English-only listings in non-English markets. Understanding why website translation is important and the full benefits of multilingual SEO is key to seeing this growth.

This makes intuitive sense. When a customer in France sees a product description in natural, well-written French (not the clunky machine translation they have learned to ignore), they trust the store more. They understand the product better. They buy more confidently.

Yet most international e-commerce stores still rely on one of these approaches:

• English only, assuming everyone reads English. This leaves conversion on the table in every non-English market.
• Google Translate widgets, which produce translations that feel robotic and inconsistent. Customers notice.
• Manual translation by freelancers, which is high quality but prohibitively slow for catalogs with hundreds or thousands of products. A freelancer charging €0.08 per word will bill thousands of euros for a 500-product catalog, and they will need weeks to deliver.

The CSV translation approach solves the speed and cost problem while maintaining quality through glossary-controlled AI translation. You keep the professional terminology consistency that freelancers provide, at the speed and price point that makes localizing a 2,000-product catalog actually viable.

Understanding your product CSV structure

Before translating anything, you need to understand what you are working with. Every e-commerce platform exports product data slightly differently, but the principles are universal.

Shopify product exports

Shopify’s product CSV is one of the most common formats. When you export from Products > Export in your Shopify admin, you get a file with roughly 25-30 columns. Here is a simplified view of the ones that matter:

The critical distinction:

Translate these columns:

• Title - what the customer sees as the product name
• Body (HTML) - the full product description, often with HTML formatting
• SEO Title - what appears in Google search results
• SEO Description - the meta description shown in search results

Ensuring these fields follow a robust SEO website translation guide will help you capture local search intent from the start. Never translate these columns:

• Handle - the URL slug (premium-leather-bag), used for routing and links
• Vendor - a brand name, usually a proper noun
• Variant SKU - your inventory reference code
• Variant Price - a number, not text
• Image Src - a URL to a hosted file
• Tags - used internally for filtering and may be referenced by your theme logic
• Type - often used for collection filtering

WooCommerce product exports

WooCommerce exports are similar in spirit but use different column names. The key translatable fields are post_title, post_excerpt (short description), post_content (full description), and the meta:_yoast_wpseo_title / meta:_yoast_wpseo_metadesc fields if you use Yoast SEO.

Custom platforms

If you are on a custom e-commerce platform, the column names will be unique to your system. The principle stays the same: identify the human-readable text fields and separate them from the machine-readable structural fields.

The translation workflow, step by step

Step 1: Export your product catalog

Export your products as CSV from your e-commerce platform. For Shopify, this is Products > Export > All products > CSV for Excel/Numbers. For WooCommerce, use a plugin like WP All Export.

Pro tip: If your catalog has more than 5,000 products, consider exporting in batches by collection or product type. This keeps each file manageable and makes quality review easier.

Step 2: Upload and choose Selected Columns mode

Upload the CSV to AI Glot. The platform will analyze the file structure automatically.

For product catalogs, Selected Columns mode is always the right choice. This mode was specifically designed for files where only a subset of columns contains translatable content.

When the mapping interface appears, mark your translatable columns and skip everything else. The explicit column mapping is what makes this approach safe: there is zero chance of accidentally translating a SKU or a price because those columns are physically excluded from the translation process.

Step 3: Set up your glossary

For e-commerce catalogs, glossaries are not optional. They are essential.

Here is why: if you are translating 500 products in the “Athletic Footwear” category, the AI might translate your brand’s “AirFlow Technology” differently across products. In one product it might say “Technologie AirFlow,” in another “Technologie de flux d’air.” A customer browsing multiple products will notice this inconsistency, and it undermines trust.

Set up glossary entries for:

• Brand names - “BrandCo” stays “BrandCo”
• Product line names - “AirFlow Series” stays “AirFlow Series”
• Technical specifications - units like “mAh” or “W” that should stay in their original form
• Material names that your brand has standardized - if you always call it “vegan leather” rather than “faux leather,” make sure the translation reflects that preference

A well-built glossary for an e-commerce catalog typically has 20 to 50 entries. The time investment is small (maybe 15 minutes to set up), but the consistency improvement across hundreds of products is significant.

Step 4: Review the cost estimate and launch

AI Glot shows you the total word count and credit cost before you commit. For a typical Shopify catalog:

• 500 products with Title + Body + SEO Title + SEO Description
• Average 80 words per product across those fields
• Total: approximately 40,000 words
• Cost: 40,000 credits in Standard mode

That is well within the Pro plan’s 40K monthly credit allowance, meaning you can translate an entire 500-product catalog for the cost of a single month’s subscription.

Compare that to a freelance translator at €0.08/word: the same job would cost €3,200 and take 2-4 weeks.

Step 5: Download and re-import

Once the translation completes, download the CSV. Open it in a spreadsheet app for a quick sanity check:

• Are the SKUs unchanged?
• Are the prices still numbers?
• Do the translated titles read naturally?
• Are your glossary terms consistent?

Then import the translated CSV back into your e-commerce platform. Shopify handles this through Products > Import. WooCommerce uses the same export/import plugin.

Handling product variants

Product variants (sizes, colors, materials) deserve special attention because they create multiple rows per product in the CSV.

In Shopify’s export format, the first row for a product contains the Title and Body, while subsequent rows for the same product contain only the variant-specific data (size, color, price). The Title and Body cells are empty for variant rows.

This means Selected Columns mode naturally handles variants correctly: it translates the Title and Body in the first row and leaves the empty cells in variant rows untouched. The variant-specific fields (Size, Color) might or might not need translation depending on your setup, so check your mapping carefully.

Handling HTML in product descriptions

Most e-commerce product descriptions contain HTML formatting: headings, bullet lists, bold text, and sometimes embedded images or videos.

AI Glot processes HTML content while preserving the markup structure. Your <ul> lists stay as <ul> lists. Your <strong> tags stay as <strong> tags. Only the visible text content inside the tags gets translated.

This is particularly important for Shopify’s Body (HTML) field, which often contains carefully formatted product descriptions with bullet points for features, bold text for key specifications, and structured layouts that took time to build.

Scaling to multiple languages

Once you have the workflow running for one target language, scaling to additional languages is straightforward.

Upload the same original CSV (not the already-translated version) and run it through AI Glot for each target language. This ensures that every language version is translated from the original source, not from a translation of a translation.

For a 500-product catalog going into 5 languages, the total effort is:

1. One CSV export (5 minutes)
2. Five AI Glot runs with the same column mapping (10 minutes each, mostly automated)
3. Five CSV imports back into your platform (5 minutes each)

Total: about 90 minutes to make 500 products available in 5 new languages. Try doing that with copy-paste.

The bottom line

Product catalog translation is one of those tasks that feels impossibly tedious until you find the right workflow. The combination of CSV export, Selected Columns mode, and glossary-enforced consistency turns a weeks-long project into an afternoon task.

The key principles to remember:

1. Map your columns carefully. Identify content fields vs. structural fields before you start.
2. Build a glossary. Even a small one dramatically improves consistency across hundreds of products.
3. Always translate from the original source. Never chain translations through intermediate languages.
4. Spot-check before re-importing. A 30-second review of the translated CSV can catch issues before they reach your store.

Ready to localize your catalog? Start with AI Glot and test the workflow on a small product batch before scaling to your full catalog.