Skip to Content - the online home of Gareth Senior

How The iTunes Charts Work

1:07p.m., Tue 27 Oct 2009

"The Charts" have always been important to the music industry. In the old days the charts were announced once a week and based on actual sales i.e. somebody physically walking into a shop and buying a single. So whichever single sold most in a given seven days was "The Number One" for the next seven days.

The chart that is announced on Radio 1 on a Sunday evening is still based on this premise - although, since 1st January 2007, it includes digital purchases.

Since iTunes has become the biggest retailer of music, the importance of the official Singles Chart has taken a nosedive. iTunes own charts have become just as important, if not more so.

But iTunes charts are much more complex. They're a rolling chart for a start. Apple haven't published any information about how their charts work, but it's my rough guess that they're based on a set of algorithms that both represent public tastes and present a chart with a similar level of stability to the official chart.

iTunes also has access to a much more rich set of statistics than purely sales. Not only can songs be downloaded, single releases can be rated and reviewed in the iTunes store. Added to which, remember that iTunes is an application for listening to and managing music as well as a retailer.

So here's breakdown of iTunes activities that may, or may not, affect the chart position of a song, album, movie, application or podcast channel.


You'd expect the most important element affecting a chart position in iTunes is the download figures. Me too. Especially for song and album charts. Podcast charts are more complicated. Speaking of which...


The key podcast charts relate to podcast channels, not episodes (although the iTunes site only offers an RSS feed of an episode chart). The success of a podcast channel can be measured using the number of downloads and/or the number of subscribers. Attracting new subscribers is a key indicator of chart success - new podcast channels that have a high visibility (i.e. a BBC Radio podcast, or just a big name attached to it) tend to chart highly and then gradually slip down the chart. Established podcasts with a high subscriber base tend to dominate the chart, even if a new episode hasn't appeared in a while.


Tracks managed in iTunes have a "play count"  - a count that is updated when the related iPod is synced with the app. PLays *could* be used as a metric in the charts, but there's a host of privacy issues involved here. If you're using the iTunes Genius feature your play counts are being processed to find new music that "you might like". It's doubtful whether play counts are communicated back to Apple and influence the charts. The Genius feature is, of course, influencing shopping habits, so it will indirectly affect chart performance by guiding new purchases. I wouldn't rule out data of this kind affecting charts in the future though. It seems to be the best measure of popularity. Of course, it might open the doors to chart rigging.

Reviews & Ratings

Do ratings affect iTunes chart performance? They could. I doubt they do since most ratings tend to be the full 5 stars - people aren't motivated to make a reasoned evaluation of a song or podcast channel and provide a genuine rating. It's the superfans that are motivated to provide a rating (see the YouTube). Reviews and ratings do influence other users to buy/subscribe, so they do, indirectly, affect chart performance.

The Editorial Affect

Apple has full power of veto over any and all podcasts, iPhone applications audio tracks and video content. Added to which, the front end of the iTunes store "features" "new" and "notable" stuff. Which must then influence subscriptions and downloads. They may even editorialise charts themselves, I don't know. Given that the charts are there to sell stuff there may be occasions where it would be in a retailer's interests to boost certain stuff in the charts - a successful first single from an album usually means the album sells as a knock-on effect. It's therefore in the interests of a retailer for these kind of releases to do better than single track releases where the fans have no more buying options.

The Smoothing Algorithm

I mentioned earlier that iTunes aim to "present a chart with a similar level of stability to the official chart". It's incredibly unlikely that downloads/sales/subscriptions directly and immediately affect the charts. These aren't "live" charts. For them to be a successful guide of influencing public taste and selling more stuff (which, for the naive amongst you, is the entire reason these charts exist) they need that social element. So User A saying to User B "have you seen who's top of the iTunes chart?" has to be a likely scenario and has to have a satisfying follow-through. So if User B goes to look at the chart as a result of a conversation the track in question has to be either still at number 1 or thereabouts.

The charts are updated regularly but each chart probably represents a bigger window than the period since the last update. A likely scenario might be that a chart is updated every hour, based on activity over the last 48 hours. Sudden spikes in sales won't immediately affect charts, but that affect will gradually come to light. The iTunes charts don't have the sudden volatility of the Radio 1 Chart (or Network Chart) where the number one track is more often than not a new release, but they probably have the same amount of change over a seven-day window, because that's the amount of change we, the public, expect.


On a technical note, iTunes charts probably aren't "live" because the process of compiling them is almost certainly a complex and expensive set of database queries and calculations. Technically, it makes more sense to do a periodical calculation.

In an attempt to prove this I tried downloading iTunes charts via their RSS feeds every minute, comparing the charts to see how often they'd been updated. This tells me nothing, since when the "updated" value in the feed changed, the actual chart didn't. This 'updated' value is just measuring some layer of the caching of chart data. If I added a cache-busting string to the end of the URL of the RSS feed I found the chart changed with every request... and then changed back for the next request (there were three instances of two tracks swapping places). So now it seems that there's some kind of load balancing going on, with two different servers serving slightly different data. Additionally, if you request a chart of 50 items (with 50 specified in the URL parameter) you don't nessecarily get 50 entries in the feed - you don't get more, some entries, I suppose, get lost in the pipes.

Clearly this is not an exact science.


Latest Posts

Blog Categories