Using ChatGPT to automate Universal’s data processing tasks
Acceleration@essenseMediacom proposed harnessing ChatGPT for streamlining manual matching processes - where movie titles in Universal’s database need to be matched with relevant information from other databases and online sources such as IMDB’s database. When providing relevant ads for a huge volume of annual global titles, seamless matching ability across all channels, and markets is key to create a holistic view of spend and performance. The initiative’s key goal was establishing a scalable way to minimise human effort required for data monitoring and cleansing; spotlighting errors such as missing film-level identifiers in the information Universal cuts its market research and marketing data by. Acceleration@essenceMediacom built a proof-of-concept app using ChatGPT to find and compare metadata with IMDB-sourced data. ChatGPT was instructed to write the code for the app, which was integrated into Universal’s Google Cloud Platform using Flask and Python. When performed manually, the task of title matching takes around five minutes per title. Considering Universal has more than 500,000 titles in its database, that is a mammoth task. Using the app, the initial time saving on a linear process is already 67%. However, with applications such as these, Universal can process multiple workflows across the different sources of data, further speeding up the process. Furthermore, Universal’s Tech and Data team is now exploring how ChatGPT can be utilised in other business units.