МТS Live ticket service and MTS AI have completed piloting the Cotype large language model (LLM)’s capabilities in generating unique SEO-optimized event descriptions for the ticket showcase.
In total, the model generated over 300 descriptions for events taking place in Moscow, St. Petersburg, Yekaterinburg, Kazan and other cities. The average search result uptick was 10 to 15 points, with some cases adding 59 or even 82 positions.
The large language model generates descriptions based on information provided to МТS Live by event organizers. Furthermore, the neural network integrates specific key queries into materials that help SEO promotion. Texts are then checked and published by humans. On average, generating and publishing a single description takes about 5 minutes, compared to 15-30 minutes a professional copywriter needs to make a text like that.
The trial project confirmed the hypothesis that the work of ticket showcase content managers can be optimized using the Cotype LLM.
МТS Live distributes AI capabilities to achieve other objectives of the entertainment business: for example, its recommendation service analyzes users’ interests and MTS big data to generate a personalized event dashboard and offer it weekly via a chat bot. The dashboard contains stage and music performances, cinema, books, etc.; the user gets a bespoke entertainment scenario that includes content from various environments. A part of МТS Live development strategy, the recommendation service has a positive impact on conversion, profit and other metrics.
