PredictHQ, the demand intelligence company, tracks 19 different kinds of events, including concerts, college calendars and sports. One of these 19 categories is Live TV Events, where a machine learning model calculates the viewership of major sports games, factoring in the teams, location, unevenness of that matchup and viewership trends and more. It is the only viewership predicting machine learning model in the world.
For 2023’s Super Bowl of the Philadelphia Eagles vs the Kansas City Chiefs in Glendale, Arizona, the model has predicted a viewership of 112.9 million ranging between 110.1 million to 115.7 million with 95% confidence interval.
PredictHQ’s CEO Campbell Brown said the model identified this figure because while the viewership was lower this year, the strength of the two teams, both number 1 in their conferences and with the 9th and 10th most fans, meant the game would draw a solid audience.
“A lot of people like to take a guess at what they think the viewership will be. Fox Sports themselves have said ~115 million. But our models, which have been more than 95% accurate since its inception, remove bias and just generate predictions based on the data. It has identified 115.7 million as the upper limit, with 112.9 million as the predicted viewership,” Mr Brown said.
In 2022, PredictHQ’s viewership prediction model achieved 96% accuracy predicting a viewership of ~117 million for the Super Bowl, and the post-game viewership was ~112 million. In 2021, they achieved 99.6% accuracy, predicting a viewership of 96 million, with a post-game viewership figure of 96.9 million.
PredictHQ’s customers use sports game viewership predictions in their demand planning and forecasting. The predictions are by county, rather than DMA (designated marketing area) to make this possible. It is especially popular with quick service restaurants and food and beverage delivery companies, which need to know which stores, warehouses and teams need to be increased ahead of each game.
PredictHQ’s event data was also used for its Predicted Event Spend algorithm, which identified the Super Bowl weekend (all Super Bowl events in Phoenix from February 9 – 12) will generate $123,780,000. This includes $78.5 million for accommodation; $12.8 million for transport and $34.8 million for restaurants. More information is available here.