A leading European shoe brand was facing considerable growth challenges in e-commerce and stationary retail. Around 2300 new articles were produced every year. However, the manual creation of product descriptions by an agency led to inefficient content management. This was reflected in the fact that around 50% of the products had no content and the existing texts were of mixed quality and were characterized by duplicate content and inconsistent wording. The time-consuming process of product copywriting meant that the company was not keeping up with product descriptions and was not as visible online as desired.

We decided to use NLG (Natural Language Generation) as an automated solution for creating the product texts. This allows a high degree of control at relatively low cost thanks to pre-defined sentence modules. Control plays a central role here, as it is reflected in the quality of the texts and thus excludes invented content (hallucinations), as known from LLMs (Large Language Models). At the same time, an NLG solution allowed us to increase the SEO relevance of the texts and define catchy opener sentences. As certain NLGs do not offer sufficient filters for text lengths, we used an open source solution to create the alt attributes (image description) and meta descriptions (texts that appear in Google searches). This gave us optimum control over the character length, which is crucial for these texts. The localization of the texts in English and French was also carried out using the NLG model. Weekly coordination and continuous development of the typesetting framework were central components of the methodology. The implementation schedule was around 2.5 months.

By using automated text creation, the company was able to increase organic traffic by 36.3% and significantly improve the quality of the content. The product descriptions are now more relevant, SEO-optimized and consistent in tone and style. This has led to increased visibility and a faster time to market for new products. At the same time, a repeatable process was set up for future products, avoiding empty text fields in the future. This strengthens the company's brand image. 

The additional translation of the texts enabled the company to improve the product pages in several markets at once. The measurable results showed a significant increase in sales.