Numbers Up For Sales

Here Comes The Science Bit

If mathematics sends your mind into a tailspin, then look away now. Because in this article I’m about to peel back the curtain on the precise numbers behind the sagging rump of the paid-for musical market. Ever wondered (as dedicated online fandoms are increasingly keen on doing) exactly how many copies the latest work by your idol has been selling over the past day or so? Unless you had friends in the music industry with access to detailed sales flashes, it was impossible to know for sure.

Only now it is. And it is all thanks to flagging sales numbers which are now exposed in a manner which surely nobody could ever have anticipated.

A Brief History Of Time

Like so many apps focused on the display and ordering of live and constantly changing data sources, iTunes is at heart an XML reader. Anyone who has ever examined the contents of their computer’s Music folder will understand this. The metadata for your entire musical database is contained in a single XML index file. This, incidentally, is the reason iTunes struggles with music collections more substantial than a couple of thousand tracks. That’s one huge file it has to scan through.

Even browsing the store part of the app works in the same manner. The client requests the live data from Apple; all delivered as an XML stream. This data was never a secret. Straightforward packet sniffing exposed just where the application was polling for its data. However the process of understanding this was made far simpler in 2012 when Apple made all the relevant URLs public and essentially opened the door for third-party developers to construct interfaces to the iTunes store or incorporate it into their personal projects.

The most interesting datasets are the ones iTunes uses to construct its internal charts. The lists of the biggest selling tracks or applications or movies of any given moment. During the height of the digital download era, the live iTunes charts were an invaluable guide to taking out much of the guesswork as to how the singles chart would look at the end of the week. Although the exact methodology used in its compilation has still never officially been confirmed, it has long been apparent that your position in the iTunes sales charts represented how many copies you had sold in the preceding 24 hours relative to everyone else.

The datasets never contain the exact sales figures, merely numbers that can be used to construct what iTunes calls the Popularity Bars, those curious blocks of lines which show you at a glance which tracks from an album or which episodes of a podcast are the most popular at that particular moment in time. The numbers are always expressed as a percentage of the most popular item in the category. Thus the Number One track of the moment always has a popularity index of 1.0, and if the second biggest was selling 92% of that total, it had an index value of 0.92.

A Valued Bookmark

Dedicated chart watchers have long made use of these numbers, thanks mainly to the work of the Dutchman known enigmatically as Kworb. His kworb.net website has for years been diligently scraping the sales data for every available country and presenting them as close as possible in an hour by hour analysis of what is selling and what is trending. But deep down it was still only possible to track the “popularity bars” and look at the underlying trends they represented. There was no easy way of determining the actual numbers behind the data. This was a neat bit of obfuscation which seemed to satisfy everyone.

However since the rise of the streaming services and the corresponding collapse of the download market this obfuscation no longer works as planned. It is now possible to calculate precisely how many copies a single track has sold in the past 24 hours. All thanks to the minimal levels of sales taking place at the lower end of the market.

Above is a screencap from Kworb, showing the state of the iTunes market on Saturday (10th March). I’ve chosen Saturday simply because this represents what is usually a high point in the market and gives us some meaningful data and above all fully representative data for us to use. I first performed this exercise using data from a Wednesday, figures which were so low they were barely believable.

Notice that the tracks at the base of the Top 100 are covered by just a few tiers of percentage figures, with several tracks tied on the same numbers. Indicating they have sold exactly the same number of copies as each other. The gap between each of these tiers is tier is also uniform – in this example roughly 0.0003 percentage points each time (subject to rounding errors given that the numbers only run to four decimal places).

The implication is clear. Every time those numbers change by such a small amount, this represents a difference in sales of a single copy. Armed with that knowledge it is a simple bit of maths to work out just how many copies each track has sold based on its stated percentage points:

x / 0.0003 (where x is the number given for each track’s popularity bars)

So, say we want to calculate the sales achieved in the last 24 hours by the Number 94 single Top Off by DJ Khaled we simply have to calculate:

200 / 3 = 66.666

We round up, of course, you cannot sell fractions of a copy after all. But yes, you read that correctly. The singles at the very bottom of the iTunes Top 100 on Saturday were selling just 67 copies every 24 hours. Using this same formula we can work our way up the table to get some more meaningful numbers. For example, how many copies per day was the Number 40 single selling?

861 / 3 = 287

Just 287 copies in a day to make the Top 40 of the live iTunes chart. So what about the Top 10? Last summer when asked I would suggest to people that about 5,000 copies a day were sufficient to make the iTunes Top 10. Is that still the case right at this moment?

3702 / 3 = 1234

That’s quite jaw-dropping. The Number 10 single on iTunes as of Saturday had sold only a shade over 1,000 copies in the preceding 24 hours. Has your online fandom propelled your latest release straight into the iTunes Top 10? Then, in all honesty, there still aren’t very many of them.

If we take this to the very top of the market, the Number One single – well that is even easier to work out. We already know that its sales are expressed as 1 – or 10000 using the multipliers I’ve been using for the sake of illustration. So, therefore:

10000 / 3 = 3333

3,000 copies a day, assuming this was consistent across the week, would amount to a total weekly sale from iTunes of the Number One single of around 21,000 copies. This number is entirely in line with reported figures for the top end of the sales market, proving that our maths here is sound.

5% Of Not Very Much

At this point, you start to see how problematic it is becoming to continually look to purchased sales of music as an adequate barometer of the popularity of the pop record. This simply isn’t the case anymore. It is merely the barometer of its popularity amongst the diminishing number of people who consume music in this manner. A month ago when Kylie Minogue released her exciting new single Dancing, there were a fair number of longtime fans figuratively losing their shit online when it failed to reach the published Top 40 chart. Noting that on pure sales terms the single was the 15th biggest seller of the week, this we were told was proof of the way the integration of streaming data has “ruined” the charts.

As we now know, becoming the Number 15 best seller of the week means very little. During her first week on sale, Kylie sold just 8,000 singles. Or to put it another way, only a shade over 1,000 copies a day spread across the week. That isn’t representative of anything approaching widespread popularity.

At the start of this year, Apple was busy angrily denying a report that it was planning the imminent sunset of the iTunes store, this despite the piece alleging this being quite specific as to the timescale it was to happen and just how the customer base was going to be transferred. Denials aside, there is without question no future in the market for purchased digital tracks, these unlikely to retain the kind of niche affection that physically pressed records and CDs still maintain. Even the numbers I’ve quoted above represent an ongoing and quite calamitous collapse in the size of the paid-for digital music market.

The iTunes track download no longer has the status of a mass-market product which it once enjoyed. Everyone has moved on, and we now have the detailed numbers to prove it.

Grateful acknowledgements to the poster ‘mjdangerous’ from the Haven music forums who was the first to indicate that these calculations were possible and whose work inspired this article.

8 Comments

  1. Just like vinyl during the 90’s and early 00’s, digital will mainly be the province of the DJ market; so it won’t completely die out.

  2. Back in the waning days of vinyl, there were often 6-way or more ties for the same position in the Canadian top 100 singles.
    It’s happening again. Songs #92 to #97 have an index of 0.0579. Then 4 more are tied at 0.0615.
    And that same 0.0037 increment repeats most of the way up the chart.
    That means the #100 song in Canada sold 13 copies in the last 24 hours. 44 copies were enough to get Dua Lipa into the top 40, 126 sales will get you into the top 10 and the #1 song on iTunes Canada sold a whole 271 copies in the last 24 hours.

  3. Whilst it’s clear download sales have fallen massively in the last few years, I think you’re being a tad dramatic. Download sales for chart hits are about the same as physical sales were around 2005.

    I agree the Kylie single for instance isn’t a genuine hit under any measure, but I’m sure I don’t need to tell you that 8,000 sales would’ve got you well into the top 40 even at the peak of the CD single.

    I’m also surprised that you haven’t been more vocal about the exact formulation of the chart because in my view it has lost its integrity, now curated playlists have such a direct impact on the charts. I have no issue with streaming be included in the charts, and there is no point in changing the weighting again to favour sales, but the way the streaming data is used as if streams and sales are the same thing is madness, and that the success of a single can often hinge on whether Spotify puts it in their Hot Hits UK playlist is an outrage.

    • I think I’ve not been vocal because I’ve nothing to be vocal about. I’m pretty content. The whole playlists thing is overstated. Since time immemorial there have been external gatekeepers for hit singles to bypass, be it getting on radio playlists, finding enough stores to stock you or being featured on the front page of iTunes. Flash back 20 years and people were bombarding me with moans about how first week discounting of new releases was skewing the charts and how it should all be banned in favour of some distant utopia which never actually existed.

      For sure, it is enormously helpful to be featured high up on the main playlists, but just as you can lead a horse to water and not make them drink, your music still has to be any good to prevent people clicking past before the 30 seconds that registers a play has elapsed. And data indicates that there is a huge amount of click-on that takes place.

      I agree with you that the existing popularity model of comparing x number of streams to the diminishing influence of a sale cannot last forever. In future we’ll move towards a dynamic points system which depends on the proportion of the market each format occupies. Complex though that may sound.

      • I don’t think it is overstated. I agree there has always been gatekeepers and the chart has never been 100% fair, but at least in the sales era, it was still an individual deciding to hand over money for a record they wanted. Yes, they can skip if they don’t like a song, but if it’s just background music to something else you’re doing you’re not really going to bother unless it’s really bad. It’s not much different from including airplay in the chart in my view. The gatekeepers are having a much more direct influence on the chart than they did before.

        A good example of a recent hit record that appeared to be really struggling on streams due to lack of playlist support was Sigrid’s Strangers. Even by the time she’d won Sound of 2018 and shot to #1 on iTunes, whilst having already spent several weeks on Radio 1 and Radio 2’s playlists, she was not even top 200 on Spotify’s daily chart. A few weeks later Spotify finally added her to some playlists and within a few days she rocketed into the daily top 40. To me, that really showed how important these playlists are. If only they’d shake it up a bit by being a little more diverse and ditching songs quicker, they could have a more positive effect on the chart at least.

        I’m interested in your dynamic points idea, could you expand on that?

  4. Also, whilst download sales for chart hits are low, the total weekly market for downloads is still over 1 million. So you can’t say everybody has moved on. A lot of people still haven’t, and some never will, because they don’t want to pay £10 a month for a service they won’t use.

    • Yep, there are the ones who hold out and strangely refuse to stream and still download. But they will only exist whilst there are still stores to them to buy from. And entities like Apple and Amazon won’t keep them running forever. Apple may deny there is a timescale for shuttering iTunes, but it will happen.

  5. It’s not that strange. If you’ve already quite happy with the collection you’ve built up, and all you really want do is download the odd single that takes your fancy now and again, maybe supplementing it with an Adele album here, a Now compilation there, you may just not feel that you need to pay £10 a month for a subscription to a service that you’re hardly ever going to use.

    I agree iTunes probably will close in the next year or two, though I think only a small proportion of those will sign up to Apple Music! It might still make financial sense for Apple.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.