We were approached by a large online electronics retailer in the USA that also sells products in many other countries. Their catalog contains over 3 million SKUs, all of which require unique descriptions.

Before automation, the process looked like this:

  • The average time to create a description for a single product (if there are photos and technical information) was about 10 minutes.
  • With at least 100,000 products, that’s over 16,000 hours of human work.
  • Additional time was spent translating descriptions, checking compliance with marketplace standards, and formatting.

The problem

  • Human resources are limited. Creating descriptions manually is a slow, expensive, and difficult process to scale.
  • The quality of descriptions is inconsistent. Texts are often duplicated, do not take into account differences between products, or do not meet the requirements of different markets.
  • International expansion is being slowed down. Without localized, relevant descriptions, it is difficult to launch new products on various platforms.

Solution: Automation with AI

The MIM:AGENCY team created a description generation system based on an AI model that takes into account:

  • technical characteristics;
  • purpose;
  • product type;
  • compatibility with other products;
  • market language.

Several AI services were considered and tested, and we selected the best one for this need.

What we did:

  1. Analyzed over 50,000 existing descriptions to identify patterns, styles, and common mistakes.
  2. We trained the model to work with nomenclature data to avoid inaccuracies in electronics designations.
  3. We configured the AI for multi-format output: short descriptions for price lists, full descriptions for product cards, and concise versions for marketplaces.
  4. We integrated automatic generation with the PIM system. As soon as a new SKU is added, the AI creates a description and submits it for approval.
  5. We added a module for automatically selecting relevant photos from an internal database or licensed open catalogs.

Results

MetricBefore AutomationAfter Automation
Time per product description~10 minutes<30 seconds
Descriptions created/month~8,000>300,000
Share of unique content70%95%+

Business effects

  • Scaling – in just a few weeks, the entire catalog was filled with product descriptions.
  • Faster time-to-market – new products appear on sale within 1-2 days instead of 1-2 weeks.
  • Reduced costs – savings on copywriting and localization were significant.
  • Quality – standardized, informative, relevant descriptions without errors or unnecessary words.

Conclusions

AI does not write better than a copywriter. But it writes faster, cheaper, and without breaks. Description automation is not about replacing people, but about effectively supporting large-scale operations. Thus, thanks to automation, the company was able to:

  • accelerate entry into new markets;
  • increase product visibility in search results;
  • standardize communication with customers;
  • reduce the workload on the team.