5,000 Product Cards in 3 Days: AI Generator for a Marketplace Seller
AI service for mass generation of SEO product cards on Wildberries and Ozon. Upload photos and specs — the system generates a title, description, bullets, and tags per each marketplace's requirements.
Challenge
The Client's Challenge
5,000+ SKUs with no proper descriptions — cards didn't rank, conversion dropped. In-house copywriters produced 25–30 cards per day: 6 months of work and a budget of ~1.5M rubles just for text. And each new shipment added hundreds of new items.
A multi-category seller operating on Wildberries and Ozon. The content task was systemic: close the accumulated catalog backlog once and get a tool for regular card updates with new shipments. Users: content managers and marketplace managers.
Solution
What We Built
We built a web service with batch catalog upload: Excel with specs or product photos → the system generates an SEO title, detailed description, benefit bullets, and tags. It accounts for Wildberries and Ozon requirements, target audience, and brand tone of voice. The manager reviews and publishes in one click.
Key Features
- Batch catalog upload via Excel or CSV
- SEO title, description, bullets, and tag generation for each product
- Adaptation to Wildberries and Ozon requirements: character limits, structure
- Tone of voice and target audience settings at category level
- Manual editing and approval before publication
- Generation history with option to regenerate individual items
Technologies
Tech Stack
Results
What the Client Achieved
Галерея
Project Screenshots
Project under NDA
Screenshots are not published under a non-disclosure agreement.
“We'd been putting off the description task for almost a year — it seemed like an endless volume. Frog Studios built a tool that closed the entire catalog in three days. Now we use it for every new shipment.”
Sergey
E-commerce Director, Client under NDA
Other Cases
Related Projects
Start a project
Ready to discuss the project?
Leave a request — we'll respond within an hour. At the first meeting we'll break down the task, show relevant cases and suggest the right entry package.