To further improve the stickiness of puzzle game players, we developed a prediction model based on the BP neural network. Firstly, we preprocessed the existing data to remove some useless information. Then, we introduced the WOA algorithm to optimize the BP neural network model and improve its prediction accuracy. To predict the distribution percentage of attempts to guess a given word in the future, we improved the BP neural network using the multi-input multi-output CIWOA algorithm (Cubic map improvement whale optimization algorithm). The experimental results showed that the WOA-BP prediction model predicted player data for March 1, 2023, to be between 20,000 and 25,000, while the actual data was 19,655. Our prediction model had a small margin of error, indicating good prediction accuracy. Based on this model, we provided reasonable adjustment suggestions for the game company’s later operations.