~30 seconds per product.
Full pipeline. Import-ready.
95percent.nl is a Dutch secondhand clothing retailer with 40+ brand partners, processing 1,500–2,500 products every week. Their catalog workflow was entirely manual. Produfoto's AI pipeline automates the entire process — one person with the app can handle any batch size, without changing a single step in their photography process.
94%
Less catalog prep time
100k+
Items cataloged
~30s
Per item
40+
Brand partners
The problem: manual cataloging doesn't scale
At 1,500–2,500 products per week, even small inefficiencies compound fast. Every product needed brand identification, size verification, condition assessment, price research, and manual data entry — all done by hand, every week.
- ✕Photography session: 2–3 hours per batch
- ✕Manual brand lookup: checking labels, searching online
- ✕Price research: Google searches per product
- ✕Data entry: typing brand, size, condition, price into spreadsheet
- ✕Quality check: reviewing and correcting errors
- ✕Total: manual — does not scale with volume
The solution: a 6-stage AI pipeline
Produfoto's pipeline takes product photos as input and returns a completed 25-field spreadsheet as output. No Photoshop, no manual lookups, no typing. The same photography session produces a fully cataloged, import-ready file.
- Photography unchanged — same photos, same time
- Produfoto reads brand from label via OCR + visual search
- Market price looked up automatically via visual search and live price intelligence
- 25 fields filled in: brand, model, size, color, material, condition, price
- Excel file output — ready to import directly into webshop
- Total: ~30 seconds per product — fully automated, web-import ready
How the pipeline works
Photo intake
5 product photos uploaded per item. SKU extracted from label via OCR.
Brand + size extraction
OCR reads the size label. 71+ brand normalization patterns identify and standardize the brand name.
Visual search
Proprietary visual search identifies the exact product — model name, colorway, retail price reference.
AI enrichment
Our AI engine extracts colors, materials, and model name from search results and photos.
Validation
AI self-correction pass: cross-checks extracted data for consistency and flags low-confidence fields.
Excel output
16-column spreadsheet generated — brand, model, size, color, material, condition, price, SKU, and 8 more fields.
Results after 12 months in production
| Metric | Before | After |
|---|---|---|
| Time per product | ~74 seconds (manual steps only) | ~30 seconds (automated) |
| Products cataloged (12 months) | — | 100,000+ |
| Brand normalization accuracy | Manual, inconsistent | 71+ patterns, automated |
| Price data | Manual Google search | Automated market lookup |
| Output format | Manually typed spreadsheet | Import-ready 16-column Excel |
| ESPR compliance fields | Not collected | All fields, every product |
A note on ESPR compliance
The EU's Ecodesign for Sustainable Products Regulation (ESPR) requires structured product data — brand, materials, condition, and more — for all fashion and textile products from January 2027. Every product 95percent.nl has cataloged with Produfoto already meets this requirement. Their catalog is ESPR-ready today, two years ahead of the mandate, with no additional work required.
Want the same result for your operation?
If you process 500+ products per week, the pipeline pays for itself in the first month. Start with a demo — upload real products and see the output before committing.
No credit card required • Results in ~30 seconds per item