Is Wordle Puzzle a True Puzzle?


(Reference: Dreamstime, 2023)

Profile Picture

Date: February 20, 2023 Name: Yiche Liu, Yiwen Ge, Xinyu Zhang

Topics: Wordle Puzzle | Trend Analysis | Data Modeling


Wordle, an online word-guessing game, has recently become a popular phenomenon with players worldwide. The game involves guessing a five-letter word within six attempts or less, and it has captured the attention of many people, with its simple yet challenging gameplay. Wordle has gained significant attention, especially after the New York Times (NYT) introduced its own version of the game, which has attracted a vast number of players.

In this report, we analyze the current version of Wordle offered by the NYT, examining its popularity, user experience, and recommendations for improvement. We aim to provide insights into the game's trends and characteristics, as well as to provide useful recommendations that can help to enhance the user experience.

Exploratory Data Analysis:

We conducted an exploratory analysis of the reported results of Wordle, using a graph to show the game's popularity peaked in February and March, after which the number of reported results started to decline, stabilizing at around 20,000-30,000. We also analyzed the distribution of hard mode, finding a maximum of about 0.137 to 0.138, a minimum of about 0.015 to 0.016, and a median of about 0.081 to 0.082 with an interquartile range of 0.061 to 0.093.

Using Spearman’s rank correlation coefficient, we found that the distribution of relevant percentages of 1, 2, 3, 4, 5, 6, 7, or more tries was significantly correlated with each other, but since the 4 tries, it showed a significantly negative correlation. For our model, we took 1-3 as negative indicators and a higher percentage on negative indicators means that the word is easier. By contrast, we took 4-X as positive indicators, and a higher percentage on positive indicators means that the word is harder.

Finally, we constructed a normalization matrix and calculated the difference of evaluation objects. We obtained the overall rating of words and classified them by tri-sectional quantiles, with some of the results shown in the file attached below

User Experience Analysis:

To understand the user experience of playing Wordle, we constructed a model that categorizes the percentage distribution of the number of attempts required to guess the solution word. For example, for the solution word "ERRIE," the distribution of the percentage is (0.23%, 3.51%, 18.33%, 31.26%, 27.50%, 15.75%, 2.86%), and we are 99% confident about it, with only 1% uncertainty. Our model enables us to classify the difficulty level of different solution words and ensure they are not too hard or too easy to guess.

We also created a decision tree based on genetic algorithm and found that the probability of the solution word "ERRIE" being easy, medium, and difficult is 0.076, 0.755, and 0.18, respectively. This tool helps discern the difficulty level of solution words and ensures they are not too challenging or too easy to guess.


We recommend that Wordle should remain free to attract more readers of the NYT and increase customer engagement to advertise other NYT games. Additionally, we suggest utilizing our model to ensure that the solution word of every day is manageable and accessible. We also recommend that NYT occasionally increases the percentage of making 5-7 or more tries to stimulate Twitter or other social media reports about 1-2 times per month.


In conclusion, Wordle is an excellent game that has gained popularity among players. To improve the user experience, Wordle should consider adding a group mode to stimulate group engagement, and set an easy-medium-hard mode to enable customers to adapt to what fits them most. We believe that our analysis and recommendations can help Wordle continue to be a popular and enjoyable game.