MediaMarktSaturn Retresco

MediaMarktSaturn automates product descriptions

More than 350,000 articles are permanently available in the online shops of MediaMarkt and Saturn. Such an enormous selection for the customer means one thing first and foremost for the members of the MediaMarktSaturn online team: texts, texts, texts. After all, every product needs a comprehensive product description, which is the only way for customers to make an informed purchase decision.

The product texts are based on information provided by the manufacturer. In the conventional process, however, it was usually only technical data, and not texts, which resulted in an inconsistent presentation of the articles in the online presence.

In addition, MediaMarktSaturn was faced with a further challenge known to all major online retailers: the more products a shop contains, the more complex the editorial creation of unique product descriptions is.


Text generation with Retresco software

MediaMarktSaturn opted for an innovative approach and fundamentally changed the process of creating product descriptions. Since 2019, the company’s online shops have been using automatic text generation with Retresco’s rtr textengine software.

rtr textengine is an application from the area of Natural Language Generation (NLG), a branch of Artificial Intelligence. Data is the foundation of the NLG application, and for texts in natural language, additional templates and conditions have to be defined in advance. Once the prerequisites have been met, the software forms unique content from data-based article characteristics such as colour, size, material and other individually-defined properties.

To date, rtr textengine has generated more than 210,000 product descriptions for articles in 427 product groups for MediaMarktSaturn. The project has not yet been completed for the online shops of MediaMarkt and Saturn. After the descriptions for all products have been composed automatically in a further intermediate step, the next major goal is to also use automatically generated texts to describe all services offered online in connection with the articles.


Great potential for e-commerce

The example of MediaMarktSaturn illustrates the potential of automatic text generation for e-commerce. The creation of product texts becomes significantly more efficient through the use of software. Natural language generation creates content in a matter of seconds and eliminates tedious coordination loops between internal players. At the same time, NLG produces content rich in variation – no product description is the same as another – a fact that is decisive for the positioning of offers in search engines.

Another plus: text generation enables the creation of consistent content landscapes. Address and tone are consistent across all product groups, which increases recognition value for the customer and thus contributes to the branding of a company.

The use of automatic text generation also changes internal content creation processes. Based on the technology, changes to large volumes of existing content can be carried out flexibly. If, for example, a new product group is added to the shop, the system generates texts for each article at the push of a button.

With this flexibility, NLG can also contribute to the success of seasonal campaigns, when special text elements supplement the product descriptions – during the Christmas season, for example.