Importing Products via CSV

# The Definitive Guide to Bulk Importing and Mapping Product Data in Folki

Last Updated: January 26, 2026

Level: Intermediate

Reading Time: 10 Minutes

While Folki offers one-click integrations for Shopify and WooCommerce, many enterprise operations run on custom ERPs, legacy PIM (Product Information Management) systems, or even simple spreadsheets. For these use cases, the CSV Import Tool is the most powerful way to onboard thousands of SKUs rapidly.

This guide is not just a "how-to"; it is a strategic playbook for data migration. We will cover the perfect CSV structure, pre-import data cleaning, handling complex arrays, and post-import enrichment strategies.

Part 1: The Anatomy of a Perfect Import File

Garbage in, garbage out. The success of Folki's AI agents depends heavily on the quality of the "Seed Data" you provide.

The "Golden Columns"

While Folki can attempt to work with minimal data, including these five columns will increase your Agent Match Rate from ~40% to 98%.

The "Context Columns"

These columns help the AI understand *what* it is looking for.

• `category`: e.g., "Electronics > Audio > Headphones". Narrowing the search space prevents the AI from looking for "Apple" the fruit instead of "Apple" the tech giant.
• `description`: Your existing marketing copy. The AI reads this to pick up clues (e.g., "features a waterproof design") that it can then attempt to verify technically.

Part 2: Pre-Import Data Hygiene (Excel Strategy)

Before you drag-and-drop your file, perform these 3 checks in Excel or Google Sheets.

1. Remove HTML Tags

If your export is from WordPress, your descriptions might be full of `<p>`, `<strong>`, and `items`.

• *Why:* While Folki can parse HTML, stripped text is processed 3x faster and costs less in token usage.
• *Fix:* Use a simple "Find & Replace" or a dedicated cleaning script to strip HTML.

2. Standardize Brand Names

Ensure "Sony", "Sony Inc.", and "SONY" are normalized to a single string: "Sony".

• *Why:* This helps Folki's agents cluster products and apply brand-specific knowledge graphs (e.g., knowing that "Bravia" is a sub-brand of "Sony").

3. Check for Duplicate IDs

Folki enforces unique IDs. If your CSV contains duplicate rows (e.g., for product variants like Size/Color), you must decide how to handle them.

• Strategy A (Parent Product): Import only the "Parent" product. Enrich it, then apply the specs to all children.
• Strategy B (Variant Specific): Ensure each variant has a unique ID (e.g., `TSHIRT-RED-L`) but include a `parent_id` column if possible for grouping (advanced usage).

Part 3: The Import Wizard Walkthrough

Navigate to Products > Import > Upload CSV to begin.

Step 1: File Parsing

Folki analyzes the first 50 rows. If it detects non-UTF-8 characters (common in Windows Excel exports), it will warn you.

• Best Practice: Always save as CSV UTF-8 (Comma delimited) (.csv).

Step 2: Intelligent Mapping

Folki's AI will attempt to guess your column headers.

• `Product Name` -> Mapped to `title`
• `Reference` -> Mapped to `id`
• `Cost` -> Ignored (Folki does not need your cost data).

The "Metafields" Mapping:

If you have custom columns (e.g., `Internal_Notes`, `Warehouse_Location`), you can map these to Metafields. The agents will not use them for research, but they will be preserved in the product object so you can view them in the dashboard.

Step 3: Validation Gate

Before committing to the database, Folki runs a "Dry Run."

• *Error Check:* "Row 412 is missing an ID."
• *Error Check:* "Row 899 has a duplicate ID with Row 12."

Correct these errors in your file and re-upload.

Part 4: Post-Import Strategy

Congratulations, you have 10,000 products in Folki. Now what? Do not click "Enrich All."

The "Test Batch" Method

1. Select 50 diverse products. Pick 5 from each major category.

2. Run Enrichment.

3. Audit the Results. Did the agent understand your specific "Part Numbers"? Did it find the right sources?

4. Tune the Settings. If the results are poor, you may need to adjust your Spec Templates (see the *Defining Spec Requirements* guide) or provide better `brand` data.

Segmentation

Use the bulk-tagging feature to organize your new imports.

• Select imported rows -> Tag as `Batch-Jan-2026`.
• This allows you to filter and manage this specific cohort later.

Troubleshooting & FAQ

My import failed with "Timeout"?

If you are importing >50,000 rows, split your CSV into chunks of 10,000. While our backend is robust, browser timeouts can occur during the upload phase on slower connections.

Can I update existing products via CSV?

Yes. If you upload a CSV with IDs that already exist in Folki, you will be asked:

• Skip: Ignore existing IDs.
• Update: Overwrite the core data (Title/Brand) but preserve any simple AI enrichments you've already paid for.
• Replace: Wipe the product and start fresh.

How do I handle multiple images?

Folki currently focuses on *data*, not media hosting. However, you can map an `image_url` column. This helps the AI (and you) visually identify the product during the review process, but checking the image pixel data is not yet a primary research method.

What about logical/boolean fields?

Folki attempts to parse "Yes", "TRUE", "1", "Y" all as `true` boolean values. Use standard formats for best results.

Conclusion: Data as Infrastructure

Think of the Import process not as a chore, but as pouring the foundation. The cleaner, more structured your input CSV, the taller and more stable your "Skyscraper" of AEO rankings can be built.

Ready to start? [Download our Template CSV](#) or go to the Import Tab.