← All case studies
Detailed case study

Sadmarket

Garden and construction equipment, spare parts, service

Sadmarket is a specialized dealer center in Kyiv, operating since 2010. Sales, warranty service and repair of garden and construction equipment. A catalog of 20,000+ SKUs with plans to scale to 100,000 and expand into European markets.

The task

Speed up catalog growth without losing content quality, and prepare the infrastructure for scaling

20K → 100K SKUs: start → target
Automated Descriptions, prices, SEO
EU Ready for expansion
01

The challenge

Sadmarket runs on Horoshop, and the platform covers the basics for now. But the business hit two problems that were slowing growth. First — a huge queue of products waiting to be added. Garden equipment isn't just the machines themselves, but hundreds of spare parts for every model. Manual card creation moved slowly, and the queue kept growing. Second — prices depend on currency rates, since most products are imported. Updating prices by hand on every rate swing meant both delays and mistakes.

On top of that, the client is especially demanding about content quality: descriptions must be accurate and relevant, attributes free of duplicates, and the site's filters must work correctly.
02

What we did

ERP sync and automatic pricing

We connected the PIM to the company's ERP — with both automatic and manual updates of prices and stock. We set up automatic currency rate tracking: when the rate changes, prices are recalculated and updated on the site with no human involved.

AI content generation with custom prompts

Regular equipment and spare parts need fundamentally different descriptions. For the spare parts categories we created dedicated prompts that account for their specifics. As a result, the system generates completely different description formats for equipment and for parts — and both meet the client's quality expectations.

  • For equipment Full descriptions focused on specifications, benefits and use cases — from the end user's point of view.
  • For spare parts Focus on the list of compatible models and replacement options with alternative part numbers.

Automatic attribute generation

The system doesn't just generate descriptions — it automatically builds structured product attributes with duplicate control.

  • Duplicate control built into the process For a site with thousands of spare parts this is critical: one duplicate attribute — and a site filter stops working correctly. Duplicate prevention tools are built into every step.

Chains of automated actions

Add a product — and the system takes it from there:

  • Generates the description With the right prompt for the category — a technical description for equipment, part-number focus for spare parts.
  • Creates attributes and SEO data Automatically, with duplicate checks and search optimization.
  • Translates and optimizes images Descriptions and meta tags — into the required language. Images — resized, converted and compressed on upload.
  • Batch processing All of these chains can be launched for whole groups of products at once.

What's next

The client plans to open stores in European countries. The system is already prepared: automatic translation of all data into the required language and synchronization with a new store — set up in minimal time.

03

The result

We automated synchronization with Horoshop — prices, stock and currency-driven changes update without manual work. Adding new product cards got dramatically faster thanks to automatic generation of attributes, descriptions and SEO. And content quality didn't drop — it improved, thanks to custom prompts for each product type and duplicate control tools.

The road from 20,000 to 100,000 SKUs is now limited not by the content team's capacity, but only by how fast the business chooses to grow.
Different products need different content. Sadmarket shows how AI generation works not from a single template, but with the specifics of every category in mind. And how a PIM can be ready for scaling long before it starts.
Status

Active partnership.

"

A large catalog with diverse products isn't a problem — if the system can generate fundamentally different content for each category.

Next case study prostor.ua Drogerie — beauty and personal care retail chain

A large catalog and high content standards?

We'll show you how AI works on your actual products