Apple’s App Charts: 2015 Data and Trends

…or how much harder it is to get into the top charts

If you’ve launched an iOS app, you know just how difficult it is to gain traction. Even when spending on marketing, the odds are stacked against your success. The iTunes ecosystem is a walled garden, where on the one hand, Apple has full control over what’s accepted, while on the other, the iTunes chart algorithm decides which apps gain visibility. These charts continuously publish and update a list of “top” applications split by categories. The higher an app is ranked in the charts, the more users see and download it. We don’t know exactly how the ranking algorithm works, but the general consensus is that both recent downloads and app usage affect it.

I’m very interested in this raking data for a number of reasons. At betaworks we invest in, build and launch many apps throughout the year. The more we understand how people use their phones, specifically which applications are used, the better we can invest, build and launch meaningful services. Since the app chart rank is somewhat dependent on actual app downloads, there’s some engagement information encoded within it.

I’ve been collecting a daily sample of the iTunes top app charts across different categories for the past two years. Last year I published a lengthy analysis that focused on algorithmic glitches — specific days which displayed abnormally high volatility in the charts.

Top plot shows days of extremely high volatility (2014 chart data), while bottom plot displays a fairly stable algorithmic system (2015 chart data)

I did not observe glitches in this year’s data. The algorithmic system governing the charts appears to have reached some level of stability. On the one hand, a stable system is great — it is predictable, there’s less uncertainty, which makes planning easier. But on the other had, there’s a stronger “rich get richer effect” — once an app makes it to an advantageous position, it’s heightened visibility reinforces it’s continued ranking.

A more stable algorithmic system means less volatility, hence a smaller number of apps that make it into the top charts.

There are two sections below. In the first, I describe the collected data and show some box plots. Then, I highlight nine interesting findings from this year’s top chart data.

The Data

Over the past two years I’ve been collecting daily samples from Apple’s published RSS feeds across multiple charts that rank iOS applications. By tracking daily chart position over time, we can observe fascinating dynamics across apps and app categories.

For example, take a look at the plot below. It compares chart position amongst a number of popular Social and Messaging apps: Facebook, Messenger, Twitter, Viber, Find My Friends, WeChat and Tango. The y-axis is chart placement: the higher the data, the lower the app’s chart position (remember, the best chart position is #1).

We don’t know exactly how Apple calculates the ranking for these charts, but we do know that it is directly affected by recent downloads. Here we can clearly see both weekly cycles in app usage, but also longer term trends throughout the year. Facebook Messenger, which relaunches as ‘Messenger’ in June, stays very close to the top position throughout the whole year, while both Viber and Tango start strong and slowly drifts down the chart. Find My Friends, on the other hand, displays high volatility — drastic changes in ranking, hence app engagement — especially throughout the summer months, and Twitter has clearly weekly cycles.

Box plots can us convey this information in a clearer fashion.

Box (Cat and Whisker) Plots

Box plots are handy data visualization tools that pack a lot of important information about the observed data distribution within their simple design. The line in the middle of each box is the median, the middle point of the data were we to sort it by value, while the box and whiskers (vertical lines above and below the box) represent the majority of observed data. The colored box is the Inter Quartile Range, a fancy name for the middle 50% observations. And finally, blue dots are outliers — data that significantly deviates from the norm.

Here are a series of box plots that represent the exact same time series data that we analyzed above. The tighter the distribution, the tighter the box and whiskers (Messenger, Facebook Messenger). The more spread out the values, the larger the difference between the top and bottom whisker (Viber). Volatile data will have outliers — blue dots — as seen with the Find My Friends plot, where the outliers represent spikes up in the charts (higher placement, lower number).

Now let’s extend our box plot to cover all iTunes classified Social Networking apps that spent at least 30 days in the Top Free App Chart in 2015. What do you find interesting here?

Social Networking apps that spent at least 30 days in the iTunes Top Free Apps chart in 2015
  • Facebook Messenger is doing slightly better than the blue app!
  • Kik is doing very well, just shy of Twitter and Pinterest.
  • Periscope is quite volatile, potentially due to certain newsworthy events that drive heightened usage.
  • Google+ is there… srsly?
  • Snapchat and Instagram are categorized as ‘Photo & Video’ apps, hence appear in the plot below:
Photo & Video apps that spent at least 30 days in the iTunes Top Free Apps chart in 2015

Other than the sheer dominance of Snapchat, Youtube and Instagram in this category, we can see that Vine stays tightly in the 50 spot, while Facebook’s Moments app shows incredibly high volatility.

If you’re interested in this data for all other Apple-defined categories, check out the following links:

Now that we’ve taken a high level look at the data, what else is observed?

1. It is HARDER to get into the Top App Charts

In 2015, a total of 1933 different applications made it into the iTunes Top Chart for Free Apps. This number is 24% less than 2014, where there were significantly more unique apps that hit the top charts, even if just for a day.

Here’s the comparative distribution of number of days in the top chart broken down by years. The leftmost vertical lines represent the top 3% of apps in terms of total days. These apps pretty much stayed on the chart throughout the whole year (think Facebook, Twitter, YouTube, and so on).

After the 3% mark, we see a fairly steep drop with a long tail. The second vertical lines represent the median for each year. This is the middle observation given all apps sorted by the total number of days they appeared in the app store. In 2014 this number was 2–1/2 days, while in 2015, it was 2 days. What’s interesting about the plot, is the difference in slope. In 2014, there were more applications that stuck around for 5–10 months, while in 2015, there are less.

If we group the data by App category, we see that across the board, there were consistently more apps that made it to the charts in 2014.

Part of this may be explained by an obvious fact: every year, new apps have to compete with a growing number of older apps. If we group the data by the application’s launch year, we see that roughly 50% of apps that appear in the top chart in a given year were also launched that year. This falls exponentially as we move back in time. But note that compared to 2014, this past year shows more of a plateau after 2013.

The fact that all of these old apps are still downloaded and used, makes it harder for new applications to get into the top charts, making it even harder to gain visibility and attract users.

Perhaps this is also due to the growing maturity and stability of the iTunes algorithmic system. As mentioned earlier, I am not seeing the glitches I observed in 2014, which suggests an overall more stable chart environment.

2. People Pay for Weather Apps!

There are a surprisingly large number of paid weather apps sustaining high ranks in the iTunes charts. The NOAA Radar Pro gets the highest chart position on average, with Dark Sky tailing closely behind.

Compare this with the free weather apps from the past year:

By plotting the change in ranking over time, versus the actual chart rank, we observe aggregate app volatility. In the case of paid weather apps, there is slightly more volatility in the Spring, yet pretty much throughout the year this market is quite competitive.

3. 2015: Summer of GIF

Something happened starting late July 2015. Even though Gif Keyboard had been around since September of 2014, usage picked up only in July 2015, around the time when Disney launched their GIF app. Other popularly used GIF apps were GIPHY and its Facebook Messenger app, as well as Giffage.

4. Sn-APP-ification

If last year’s big trend was to add ‘flappy’ in your app’s name, this year it has been replaced by ‘snap’. There are numerous ‘snap’ apps that sustain their rankings in the app store. Many of these apps, as you can see below, help users snap better — take videos, upload, edit, download, save, etc.

5. Instagram: Repost, Layout, Followers

The wildly popular Instagram app has stayed well within the top 10 apps throughout 2015. Following the popular trend, team Instagram also launched their own Layout app, which has performed phenomenally well. There are various other layout and repost apps for Instagram, but the one app that consistently stayed in the charts is Followers +, which gives users the ability to gauge their IG followers.

6. Notable App Launches in 2015

Here we look at new applications that not only reach a notable spot in the top chart, but also sustain it to some degree. As you can see below, a handful of these apps are music related: SongFlip (free music streaming), Musicloud (stream music from your Dropbox mp3's), and Free Music HQ (what it sounds). Moments is the Facebook app that helps you find yourself in friends’ photos, and Triller helps users make music videos on their phones. There are also two applications that help upload content into Snapchat. See a trend here? Media, media, and more rich media!

7. Volatility

Applications that display volatile ranking typically engender services that have varying usage over time. Some of these volatile apps stay in the top charts throughout the year, yet continuously shift up and down. These consist of the NFL and ESPN apps, which spike on days of the week when games are on. Others, such as — CVS, Dunkin’ Donuts, and Fox Now — perhaps vary due to promotions or special deals. The Dunkin’ Donuts app tends to fall mid-week, while the FOX NOW app dips in ranking towards the end of the week.

8. Less Volatility for the Bible!

In 2014 we saw large fluctuation in Bible app usage, especially towards the Summer. Alternatively, in 2015, the Bible’s rank is significantly less volatile, where, unsurprisingly, peak usage and downloads happen on Sundays. Christmas and Easter are slightly higher compared to the rest of the year. Shocking, I know.

9. And Finally, the Kardashians

While Kim had an older iOS app earlier in the year, the three siblings had a coordinated launch of their new apps on September 14th, 2015. The data speaks for itself — Kylie Jenner is clearly on top.

I could go on and on here, but will cap this already too long blog post at this point. The iTunes dataset is so interesting as it is one of the only ways in which we get a peek into usage of applications on iOS devices. This is helpful when building and launching iOS applications.

But it is also important to track this type of data, and other ranking systems, as a way to hold these companies accountable.

There’s true power encoded within the algorithmic system governing these charts. If your app makes it into an advantageous position, you will benefit wildly.

We invest time and resources into the iTunes ecosystem, yet are governed by an increasingly opaque algorithmic system. By studying inputs and outputs, as Nick Diakopoulos suggests, we have the ability to monitor and keep these systems in check.