2023 Disney World Crowd Calendar Retrospective

If you were paying attention last week, you know that I celebrated the new year by comparing Touring Plans predictions to Disney’s ticket costs “prediction” about crowd levels. The results were pretty favorable to Touring Plans, and you can catch up with them here. This week I’m digging a little deeper and trying to determine how crowd calendar predictions performed compared to previous years and at the park level. All of this will help the stats team to make even better predictions in the future, and to help you make better plans throughout the 2024.

Explain the Math!

If you take a peek at our Walt Disney World crowd calendar, you’ll notice that each park-day is given a predicted crowd level on a scale from 1 to 10. Crowd level 1 days are the least crowded of the year, and crowd level 10 days are wall-to-wall humans everywhere you look. These are Touring Plans’ predicted crowd levels.

Then, after a park day happens, we can measure all of the wait times at every attraction in every park and determine what the actual crowd level was for that day. By subtracting the difference between the actual and the predicted crowd level, we can get an easy measure of how accurate those predictions were.

In this case, I’m going to subtract the predicted crowd level from the actual crowd level. For example, if Touring Plans predicted a crowd level 10 at Magic Kingdom, but the day ends up being a crowd level 8, the difference is -2. The crowds were overpredicted by 2. But if Touring Plans predicts a crowd level 4 at EPCOT and the day ends up being a crowd level 7, the difference is 3. The crowds were underpredicted by 3.

In general, you’re not going to feel any difference in your park day if predictions are off by just one crowd level. Within 1 crowd level is the goal. Really anything within 2 crowd levels I’ll call “fine”. Missing by 3 or more crowd levels is a big miss, and that’s the type of thing that should be avoided.

2023 Performance Compared to Previous Years

In an ideal world, that green bar would be 100% every year. Touring Plans would have perfect predictions, you’d love and believe them 100% of the time, and we’d all be best friends. In reality, things happen. The world is an unpredictable place. More to the point, Walt Disney World can change their operational decisions on a whim. Capacities, staffing, hours, etc. are all under their control. And the stats team tries to not change the predicted crowd levels constantly so that people planning have some stability. These operational changes have shifted into hyper-drive post pandemic, along with the unpredictability of human behavior. So there’s a lot working against stable predictions. Still, everyone wants that green bar in the graph to be as big as possible. And we especially want the red and pink bars (missed by 3 or more crowd levels) to be as small as possible.

What about the other bars? Well, in general, people are happier if the parks are less crowded than they expect (aka, Touring Plans overpredicted). And they’re more upset if the parks are more crowded than they expect (aka, Touring Plans underpredicted). But Touring Plans is full of math nerds, and math nerds want our distribution to be centered (aka, just as many underpredictions as overpredictions). Still, during a retrospective, I’m going to recognize and acknowledge that all of you are going to be less angry if those underpredictions are kept to a minimum.

2023 wasn’t bad. In some ways, it was the best prediction year since 2020. And that year, Touring Plans could predict a 1 and the actual crowds would be a 1 most of the time after reopening. The comparison to 2019 is more interesting. And that green bar is significantly larger in 2023 than it was in 2019. Unfortunately, the location of that green bar in 2023 is more problematic than in 2019 … or any year since. In 2019, Touring Plans underpredicted by 2 or more crowd levels just 18% of the time. In 2023, that number shrank to a barely noticeable 2% of the time. On the one hand – yay! Very few times that anyone was angry about unexpectedly large crowds. On the other hand, Touring Plans overpredicted 14x more than we underpredicted, so things are more skewed than they’ve ever been.

2023 Performance Compared Across Parks

In order to understand what went right (or wrong) in 2023, we need a little more information. In this article we’ll go down one level and look at how predictions performed at each park rather than the resort as a whole.

We can quickly see that Magic Kingdom is the park that the predictions got “most right” in 2023. Touring Plans predictions were within 1 crowd level 61% of the time. And Hollywood Studios is where there was the most trouble in 2023 – predictions were only within 1 crowd level 56% of the time.

Magic Kingdom wins again for least “skewed” predictions. At that park in 2023, Touring Plans overpredicted by 2 or more crowd levels 20% of the time, and underpredicted by 2 or more crowd levels 18% of the time. Nice and centered. At Hollywood Studios, we see the most “skewed” predictions. Touring Plans only overpredicted by 2 or more crowd levels 2% of the time. On the other hand, we underpredicted by 2 or more crowd levels 41% of the time. 20x more underpredicting than overpredicting is a big ouch. At least it means that Hollywood Studios was more pleasant this year that it was supposed to be!

Calendar Retrospective of Crowd Calendars

What better way to review calendars than with more calendars?!

I’ll start here with the park where predictions had the highest accuracy, provide my commentary, and then move to the next park. We can discuss park-specific issues, as well as trends that I notice affecting multiple parks.

Magic Kingdom

At Magic Kingdom, predictions were within 1 crowd level 61% of the time. This is represented by all of those lovely white days on the calendar. When Touring Plans overpredicted crowds, those days turn purple. And when predictions make people angry by underpredicting crowds, those days turn orange. The darker orange, the angrier the people.

We start off with an interesting January. Overall, late January was less crowded than expected. This continues a trend that we saw in 2022, but that dip we thought was due to Omicron. Perhaps instead it’s the beginning of a new regular drop in crowds. Most days in February and March were less crowded than expected too.

In April there was one week that had more Spring Break visitors than predicted, but then we hit a glorious uncrowded May. Throughout the summer, weekends were less crowded than expected. This fits with our observations that in a summer of record-breaking heat, locals stayed home instead of hitting the parks on the weekends.

In the fall we see the return of my favorite crowd trend – “Magic Kingdom party day behavior” On party days, the Magic Kingdom closes early so that party guests can enjoy the more-empty-ish park that they paid to experience. That means that regular day guests tend to avoid these days and instead go to Magic Kingdom on other surrounding days that have longer operating hours. Which makes sense, if you don’t think about ALL of the other humans making the same decision. What ends up happening is that party days have much lower crowds, and non-party days that are sandwiched in-between two party days have much higher crowds. We know this phenomenon exists, and the crowd calendar got it wrong a lot in 2022. But it continued to get it wrong in 2023 too.

Here’s a visualization of the problem. On non-party days, Touring Plans tended to under-estimate the crowds by an average of almost 1.5 crowd levels. Those lines on either side of the box are the 95% confidence interval based on the sample we have. On party days, on the other hand, Touring Plans tended to over-estimate the crowds by an average of 1.1 crowd levels. These results are almost identical to what we saw in 2022.

Animal Kingdom

Animal Kingdom still has a lot of white-colored days. And late January looks like Magic Kingdom with its overpredicted days. But then in February, Animal Kingdom got a random crowded week! For a solid 7 days, crowds were at least 2 crowd levels higher than expected. And then in April, Animal Kingdom’s crowded week came 7 days earlier than Magic Kingdom’s crowded week.

Not just weekends were uncrowded during the summer in Animal Kingdom – in fact most of June and July were well below expectations. Things evened out in July, but then turned toward lower crowds than expected for pretty much the rest of the year.

2023 trends here are interesting. In 2022, overpredictions at Animal Kingdom mostly happened during cooler months in Orlando. In 2023, Animal Kingdom was just less crowded throughout the whole year … other than February.

EPCOT

In 2022, right when the Festival of the Arts kicked off, crowds begin heading to EPCOT at a rate that was much higher than predicted. This year the predictions overcorrected and expected more people than actually showed up. And those crowds stayed away for a significant portion of the year, especially in late April through mid-July.

But then in September, EPCOT had its day. All kinds of things were going on in EPCOT in September that made it relatively more exciting than the other parks – Soarin’ Around the World returned, Journey of Water made its debut, and Figment … fell on his face a few times. EPCOT is always popular with locals on the weekends, but it became especially popular on Sundays and Mondays in September, and those represent a very large portion of the all of the underpredictions made at any park for the year.

Hollywood Studios

Ah, Hollywood Studios. Remember, this was the most “skewed” park, and we can see it in the coloring here. It was the most skewed last year too – except last year it was all underpredictions and this year it’s all overpredictions. The crowd calendar overcorrected!

This calendar also has our biggest miss of the year – an overprediction of 7 crowd levels on December 16th. That day was a total washout on what otherwise would’ve been a crowded holiday-season weekend.

January, February, July and August were the least-bad, with a few scattered underpredictions to balance out the overpredictions from the rest of the year.

 

Did any of the results of this retrospective surprise you? Were you at Walt Disney World for any of the significant overpredictions or underpredictions? Let me know in the comments!