Feast or famine: Data during the holidays - Mixpanel
Product Foundations

Feast or famine: Data during the holidays

Parker Tarun

A week ago, Americans took time off for a particularly American holiday: Thanksgiving. They gathered around the dinner table. They looked their loved ones in the eyes and expressed gratitude. And then, they took refuge in their smartphones.

That’s what billions of Mixpanel data points tell us, anyway. The average number of actions per device daily is 13.66. On Thanksgiving, that experienced a slight uptick to 14.88.  That itself isn’t much. Americans were relaxing (or perhaps avoiding another political discussion), and so they got on their devices. Where things get interesting is when you slice and dice the data.

Looking back at the Thanksgiving holiday, we dove into our Mixpanel Trends data to observe what users were up to. Usage patterns on apps and websites didn’t just increase; they changed in meaningful ways: 

  • The most engaged users over the holidays aren’t necessarily who you’d assume,
  • The types of apps that get love are radically reordered.
  • And e-commerce not only reflects our shopping patterns, but begins to change them. 

Whatever kind of product you’re building the long weekend might not be the time to tune out.

A nation divided—though not how you think

Conventional wisdom would have it that urbanites can’t get off their phones and computers. The data tells a very different story. During Thanksgiving, Americans in rural states performed a far greater number of actions on their phones on average.

Top 5 Most-Distracted States

  1. Wyoming – 24.94 actions (up 73% from 30-day average)
  2. South Dakota – 24.09 (up 59% from 30-day average)
  3. Vermont – 21.24 (up 45% from 30-day average)
  4. Utah – 18.12 (up 59% from 30-day average)
  5. Montana – 17.62 (up 59% from 30-day average)

It’s not just that rural America was distracted either. City slickers were actually off  their devices. For comparison, tech-savvy Californians were 9.4% less engaged on their apps during the holiday than average.


Why does any of this matter?

Product builders often premiere apps in crowded urban markets early. Sometimes, with good reason. But, as our iOS 10 data revealed earlier this year, there is often more engagement in rural areas. So, if engagement is key (and driver fleets aren’t) maybe your app should beta in Wyoming.

And now we turn to another question: If Americans overall weren’t socializing with family members, what were they doing?

Turkey slump

Thanksgiving 2016 was a sad day for developers trying to keep users healthy. Or educated. But it was a pretty good day for anyone just trying to have fun.

Apps and Website Engagement on Thanksgiving

  1. Photo/Video (up 26.7% from the day before)
  2. Gaming (up 26.25% from the day before)
  3. Health (down 1.4% from the day before)
  4. Social (down 13.5% from the day before)
  5. Education (down 24.6% from the day before)

As the percentages show, usage patterns really do change during the holiday. On The Signal, we’ve harped on the idea that product builders need to have empathy for their users. This holiday season is no different.

For example, a fitness app that shall remain nameless sent me a push notification to “get moving” on Thanksgiving Day. This was likely triggered by my recent inactivity, but you can understand why it resonated as tone-deaf. I’m not thinking about going for a run on Turkey Day.

All things considered, though, the health category’s dip was relatively small. Education surely took a hit, but what surprised me most was social. Perhaps Americans were too busy interacting with real family and friends to pay attention to their virtual counterparts. But that isn’t to say they weren’t on their phones.

Photo and video app engagement shot up on Thanksgiving Day, as did gaming. These increases might seem obvious, but they shouldn’t be ignored. A boost of over 25% represents a major opportunity for product folks. Those working in gaming or photo/video-sharing should pay attention to this trend. Users have downtime. They’re eager to have fun. Capitalize on it. After all, not everything’s about e-commerce.

But a lot of it’s still about e-commerce

Black Friday came early this year. In 2015, shopping and e-commerce apps and sites experienced a 5.2% decrease in actions on Thanksgiving Day. This year, it flipped. And then some. E-commerce and shopping products had a 10.8% increase in actions on Thanksgiving day.

So, what happened? Did the sales start earlier? I’m not sure. Whatever the case may be, we now know consumers don’t have Black Friday fatigue. Over the past few years, shopping madness has extended into Cyber Monday, but this data suggests that online consumption is creeping the other way, too.

There is an audience for earlier deals. The increase on Black Friday eve wasn’t a fluke either, because the pattern repeated itself on Sunday. Cyber Monday hadn’t even struck yet, but Sunday’s e-commerce and shopping actions were up by 14.52% from Saturday (the weekend’s low point).  


Looking at these numbers, it seems silly to confine virtual retail to a single day like Cyber Monday. As more of the market moves out of big box stores and onto the web, discounts deserve a less rigid cadence. Users are looking to buy throughout the weekend, starting on Thanksgiving. E-commerce and shopping products would do well to heed this trend. Cyber Monday might eventually give way to Cyber Week.

Feast or famine

We all want to believe users are opening our apps every day and using them regularly. That’s naive. Products go through booms and busts, but these high- and low-use periods don’t have to be mysterious. When you can anticipate high-use periods, such as a holiday, your team can collect and analyze a lot of good data.

In the coming weeks, it would behoove builders to think about how their products can ride the waves of engagement. A 27% increase in engagement isn’t just something to celebrate; it’s something to capitalize on.

As for the health apps?

Well, January should be fruitful.


Get the latest from Mixpanel
This field is required.