TV measurement incorporates Artificial Intelligence to predict audiences
Television audience measurement is strengthening its commitment to Artificial Intelligence in order to obtain more detailed information on viewer consumption. One of the consultancy firms that has redoubled its efforts in this area is GECA, which, together with Microsoft and its partner Plain Concepts, has just presented the first tool that predicts television audiences 30 days ahead.
This consultancy firm, a subsidiary of Mediapro, has been applying this audience prediction exercise for a decade, albeit with a very limited level of precision, obtained manually and whose results had to be read by several analysts. Three obstacles that the new tool aims to combat through algorithms, after a long work developed by several professionals including mathematicians, experts in new technologies, R&D and television.
The general manager of GECA, Kike Lozano, explains that this tool “serves to improve in terms of efficiency the management of commercial inventory in television and to maximise profits by minimising the impact on programming proposals”.
One of the innate complexities of this audience prediction model is the multiple factors – such as the weather, for example – that can influence television consumption and, therefore, significantly alter the forecasts, which means that human work must continue to be used.
The team working on the tool, Lozano explains, “we paint grids in which we foresee that certain slots will be broadcast. From there, we assign tags that, as this information is updated, such as the interviewees for a programme or the films that are broadcast, allow us to pinpoint audiences with precision thanks to automatic learning”.
The consultancy describes the system as being able to analyse a range of factors such as demographics, viewing preferences, social trends, historical data and content minutes.
