CatalogFix AI — Supplier Catalog to Clean Product Data

Clean supplier catalogs. Nothing invented, nothing guessed.

CatalogFix AI converts messy supplier PDFs, Excel files, and CSVs into structured product data with a clear Ready / Needs Review split.

It is built for one thing most extraction tools handle badly: uncertainty.

If a supplier SKU is not printed or cannot be verified, CatalogFix does not fabricate one. If a file is a technical datasheet rather than a product catalog, CatalogFix can intentionally return zero product rows instead of turning technical references into fake products.

Best for

• supplier catalogs that need to become e-commerce product data;
• image-only PDF catalogs that require OCR;
• mixed catalogs with printed and unprinted supplier codes;
• price lists and order forms;
• teams that need an audit trail instead of a black-box extraction.

What you get

Every successful catalog run can produce:

• RESULT.xlsx — full workbook with Clean Master, Issues Found, Shopify Ready, Needs Review, and Import Report;
• SHOPIFY_READY.csv — only rows that passed the current release gates;
• SUMMARY.json — counts, document type, and quality statistics;
• IMPORT_REPORT.json — page/sheet routing and parsing audit;
• Dataset — normalized product rows with source and quality context.

Ready vs Review

CatalogFix does not treat every extracted row as equally trustworthy.

Ready

A row is placed in Ready only when it passes the current release gates, including required product fields and quality checks.

Needs Review

A row is placed in Needs Review when something important is uncertain, for example:

• supplier SKU is missing or not confidently verified;
• OCR confidence is low;
• title or product-card boundaries are uncertain;
• required commercial data such as price is missing.

This means a run can legitimately return zero Ready rows while still extracting useful product candidates. That is intentional behavior, not a failure.

Safety behavior

CatalogFix prefers an explicit review state over false certainty.

• Missing supplier SKUs are not invented.
• Internal candidate IDs stay review-only.
• Technical datasheets can be rejected with zero product rows.
• Visual OCR keeps source location and confidence metadata.
• Structured price sources preserve price provenance where available.
• Quality gates keep uncertain records out of Shopify Ready.

Knowing when not to extract is part of the product.

Input

Upload one supplier catalog per run.

Supported formats:

• PDF
• CSV
• XLSX
• XLS

You can also provide a direct HTTP(S) URL.

For URLs that do not preserve a recognizable file extension, use Filename override.

Verified release behavior

CatalogFix AI v1.9.0 was regression-tested across:

• image-only visual catalogs;
• structured dual-price order forms;
• technical datasheet refusal.

In the image-only Appliances test, the Actor successfully ran RapidOCR on Apify, extracted structured product rows, preserved verified supplier codes such as DW-01, MWO-1, FSCR01, BIO-01, CW-165, and CW-46, and kept uncertain rows in Review.

In the TL972 technical-datasheet test, CatalogFix intentionally returned zero products with a technical-datasheet skip status.

How it works

1. Upload a catalog.
2. CatalogFix classifies the document and routes pages by content type.
3. Structured parsers handle commercial tables and order forms where possible.
4. Visual pages use adaptive OCR.
5. Quality gates separate Ready rows from review-only rows.
6. Download the structured dataset and audit files.

Pricing

CatalogFix uses pay per event with one charge per completed run:

• Up to 50 pages: $4.90
• 51–200 pages: $12.90
• 201–500 pages: $24.90
• Technical/statistical non-catalog screening: $1.00

The current self-service Store version accepts PDFs up to 500 pages. Larger catalogs should be split or handled as a managed/custom run.

Notes and limitations

• OCR-heavy PDFs take longer than text-based catalogs.
• A second OCR pass may be triggered when the first pass is not reliable enough.
• A real supplier SKU can still remain in Needs Review if other required fields are missing.
• CatalogFix does not promise perfect extraction. It makes uncertainty explicit and auditable.

Current release

CatalogFix AI v1.9.0