Which albums had the highest number of worldwide sales?

As has been mentioned elsewhere this site focus on success in the source charts, not on estimates of sales. In our experience most quoted sales figures are inherently untrustworthy and should normally be treated with extreme caution. However there are some music industry based organisations that are responsible for defining official "achievement levels" (like Gold and Platinum sales for example). The audits of these organisations can be trusted as evidence of sales numbers and from these we can calculate a guaranteed minimum number of sales for each album that is audited (in millions). Of course many albums have not been analysed in this way, their absence from these lists indicates only that no-one has paid for them to be audited, not that they didn't sell the required number of copies. In particular albums from the 1950s or 1960s are noticeably absent, even though a historical review would appear to suggest they should have sufficient sales. There are a number of other issues that affect these numbers, they are described in detail following the table.

This table also includes a calculation, that is our estimate of the most probable level of worldwide sales. Like all such estimates this number should be treated with some caution, we have also included a range of possible values to indicate the uncertainty of the figure. For example, our estimate of the sales of "Thriller" are in the range 48-65 million.

Number of copies

Of course the numbers indicated in this table are for legitimate sales, they provide no details of the number of actual copies out there. As an example during the 1960s and 1970s the sale of music by The Beatles was illegal in the Soviet Union, as a consequence there was a thriving black market in bootleg copies of their albums and copied photographs. Even when these restrictions were relaxed in the early 1980s the local music publishers sold enough of their music for at least one current record company to claim that it was originally financed by the unauthorised sales of Beatles music. Until the mid 1980s there was no attempt to place this business on a legal footing.

Looking at the music revenue as a percentage of GDP (as discussed here) suggests that China, Brazil, Russia, Mexico and possibly Italy all sell significant numbers of albums without reporting them, these sales are not included in the above figures. The numbers here indicate approved sales, not the number of copies that exist. Of course it is difficult to guess which artists and albums are most involved.

Organisations

The four organisations that have been used are the official bodies in the four biggest relevant markets. Each of these groups issues the sales certificates for a particular country. They all will only audit album sales if a record company both requests and pays for the work, which means that only albums that someone is interested in promoting get examined.

RIAA for the USA

The "Record Industry Association of America" certifies albums sold in the USA. A level of "Gold" indicates 0.5 million sales, "Platinum" 1 million sales, "2 x Platinum" 2 million sales and so on. The information shown here is from www.riaa.com/goldandplatinumdata.php.

There is some confusion in the sources about the way that RIAA counts double albums. Sources indicate that sales of most double albums are counted as two records, so selling 5.5 million copies of, for example, "Speakerboxx & The Love Below" (a double CD) results in a "11 x Platinum" award. However some sources clarify this as only being true for records of longer than a certain duration and state that, for example "The Wall" was short enough for its "23 X Platinum" to indicate 23 million sales. In addition the packaging can vary between vinyl and CD, so for example "Grease: The Original Soundtrack" was a double LP (all sales counting twice) but only a single CD. For this table we have elected to just ignore all these issues.

BPI for the UK

The "British Phonographic Industry Ltd" certifies albums sold in the UK. The levels listed here are "Silver" 60,000 sales, "Gold" 100,000 sales, "Platinum" 300,000 sales and "Multi-Platinum" which indicate multiples of 300,000 sales. This data can be examined at www.bpi.co.uk/platinum/platinum.html.

SNEP for France

The "Syndicat National de l' Edition Phonographique" certifies albums sold in France. The levels listed are "Gold" 75,000 sales, "Platinum" 200,000 sales and "Diamond" 750,000 sales. This data comes from www.disqueenfrance.com/.

BMi for Germany

"Bundesverband Musikindustrie" has certified albums sold in Germany since 1975. The levels are "Gold" 100,000 sales and "Platinum" 200,000 sales, but then rather oddly "3 X Gold" 300,000 sales and "2 x Platinum" 400,000 sales and so on. There are also some albums certified as "10 X Gold" (rather than "5 x Platinum" for some reason), our guess is that has to do with the logical way that Germans think. This data came from www.musikindustrie.de/gold_platin_datenb ank.html, unfortunately a really bad web site design makes it quite difficult to discover album certifications especially for groups towards the end of the alphabet for any year after 1995.

World Market

You will have noticed that the definition of, for example a "Gold" record varies from country to country, this is because it is tied to the anticipated sales in each country. When considering the possible global sales of albums the relative size of the market in different countries has to be taken into account. The IFPI list music revenue from the top 30 countries at www.ifpi.org/content/section_statistics/ index.html. This shows that the distribution of sales in 2003 was:

The album sales figures for the US, UK, France and Germany have been calculated from the certification levels. The Japanese market is 15% of the global one, however it is dominated by local music that does not sell outside the country. For this reason the Japanese market for international albums is actually less than 2% of the World market. The other markets also have significant local content, indeed once that is taken into account the four "regions" above each represent about the same size of adjusted global market. In other words the expected sales of an album should be 7x in the US, 2x in the UK and 1x each in France, Germany, Southern Europe, Northern Europe, Anglophone and everywhere else. That is the four major markets should be 11/15 (that is 73%) of the total.

Even without this adjustment the four largest markets represent 68% (that is about 2/3rds) of the world's market. So if, for example, "Thriller" sold 40 million copies in those four markets then the expected total global sales would be 60 million. If someone were to claim global sales of, for example 100 million, that would require the rest of the world to have purchased 60 million copies, or 3 times the anticipate sales (and that's already assuming the album is the best selling one ever). For those sales levels to be correct the album would have to have been 3 times as successful as its nearest rival, which of course would have to show up in the charts of the majority of other countries (and clearly doesn't). So a claim of more than 100 million sales for "Thriller" would be obviously ludicrous.

By using the 68% figure (rather than 73%) we are overestimating the sales, however since we are doing so for all albums we will stay with this figure.

Calculating the Probable sales

Sales over time

As was mentioned the certification authorities started work in the 1970s. This means that reliable sales numbers are not available for albums released before about 1980. As other pages have shown the impact on the chart of, for example some albums from The Beatles, was massive. Yet albums like "Rubber Soul", "Revolver" and "Let it Be" are not even in the above list. Can that possibly be accurate, or has the focus on recent sales resulted in a systematic underestimate of historical sales?

We can start to explore the question of historical sales levels with a set of numbers from the RIAA for the US music market over the period 1990-2005. This illustrates how the US revenue from music has changed over the last 15 years, however this series of data only starts in 1990, much too late to answer our question.

Here is another indicator, the number of Gold disks awarded by the RIAA since 1955. One way to view this says that the number of certified disks in 1970 was roughly half that in 1990 so the ratio of sales should be about the same. Awards are always retrospective, so one might worry that this assumption is too simplistic. However it is reasonable to assume that the rate of new awards in any year is related to the number of candidates passing the threshold for the award within that year, in other words the the number of awards is proportional to sales in that year, even if the bulk of the album's sales were obtained in previous years. In fact it is easy to see that this measure will exagerate the volume of sales as time goes on, since the number of candidates for awards is always being added to.

So the sales for 1990-2005 and the number of gold records both help to hint at the trend in market size. Using a combination of other such indicators we can estimate the music market sales for the US for a much longer period.

We can make some reasonable assumptions about how the balance of music sales changed from the US led music scene of the 1950s to today's more global market (with the US accounting for just over 1/3 of the sales). This results in the above estimate of the value of the global music market from 1955-2007.

Finally, if inflation and the changing cost of albums is taken into account the above plot shows how the real global music revenue has changed over the last 50 years. There are two factors that complicate the interpretation of the peak between 1970 and 1980, first this era was dominated by the sale of singles rather than albums, and second the adjustment to account for the relative cost of albums may be too severe.

Even taking these factors into account it is reasonable to suggest that sales of top hit albums in the late 1960s were at least 50% of those in the 1980s. So we should expect that a "major hit" album of 1965, such as "Rubber Soul" would have something like half the sales of an equivalent album from 1995, like "Jagged Little Pill", especially given the 30 years worth of additional sales. However the estimated sales of "Rubber Soul" are less than a quarter of "Jagged Little Pill". This appears to indicate that the low representation in this sales lists of The Beatles (and of course all the other top acts of the 50s, 60s and early 70s) are connected with the process of auditing rather than reflecting the actual sales numbers of their records.

Of course it is interesting to compare this analysis of the global sales with the long term trends that can be seen in the charts.

As with all the complex calculations described on the site you can decide to try out a different approach, the available CSV File gives you all the data you need. If your analysis shows something interesting we will be interested to hear about it.

Source Data

These results were generated from data version 1.8.0065 of the data.

Back to Introduction

Previous Comments

2 Dec 2009

Genesis biggest seller

Hi I read with a friend of mine on a website that Genesis highest selling album is not We Can't Dance it is actually Invisible Touch. It looks like to me that your numbers and maybe statistics are outdated. We Can't Dance sold 22,000,000 copies worldwide and went 4 times platinum in the U.S. and Invisible Touch went 6 times platinum in the U.S. and has sold 24,000,000 copies globally. I even read on the VH1 website that Invisible Touch is Genesis highest selling pop album to date.

Thanks Aaron Wake

First of all thank you for your input. Of course most of the lists on this site measure success in the charts round the world and not sales. So, for example, the 1983 album "Genesis" lists higher than "Invisible Touch" because it was a hit in more countries, but it is fairly certain that its sales were lower.

Claims of worldwide sales on web sites are generally not trustworthy, for example on Wikipedia look at the number of page modifications and the length of discussions on all pages that mention sales numbers. The issue is that there are no validated worldwide sales numbers so every group of fans use their own approach to estimate numbers.

We do have a page that lists our estimates of sales numbers. We use a combination of the certification levels from the US, UK, France and Germany (the four biggest markets). That page describes how we combine those numbers to estimate worldwide sales. It also explains why we don't trust even our own listing of sales numbers.

The two albums you mention are listed on that page. Your numbers for the US sales are correct, however you don't list the fact that "We Can't Dance" was also 5xPlatinum in the UK and 5xPlatinum in Germany, while "Invisible Touch" was 4xPlatinum in the UK and 1xPlatinum in Germany. "Invisible Touch" was more successful in the US, however the certifications and chart positions show that "We Can't Dance" was clearly a bigger hit in Europe.

Our estimate is that "We Can't Dance" sold 15-24 million worldwide and "Invisible Touch" sold 12-17 million.

We don't know where VH1 got these numbers from, however we have seen other occasions when VH1 has over emphasised acts that were successful in the US and ignored acts that were hits anywhere else in the world. This is one of the reasons why we don't use any of their charts in this site.

So, we believe that the balance of evidence suggests that "We Can't Dance" sold more, however the wide margin of error leaves the question open. We also have learnt that you should not trust any website that lists worldwide album sales numbers (even our own).

30 Dec 2009

LAUNDRY SERVICE SHAKIRA

Where's "Laundry Service" by Shakira? It sold 20 million copies.

Looking at the album page for Shakira we can see that it was certified as US 3 x Platinum, UK 2 x Platinum, France 2 x Platinum, Germany 2 x Platinum.

Which means it had sales of at least 4.4 million, and by our calculations probably sold 12.8 million (with an estimated range of 10-16 million sales).

By these numbers the album should be on this list. However this English language album was originally known as "Servicio De Lavanderia" in France, so when the list was calculated the French numbers were omited and the resulting estimate of 9.9 million sales was below the cut-off. This mismatch has now been fixed, so next time list is calculated it will be included.

Sales of 20 million are much higher than our estimate. And the fact that both BMI and Shakira's own web site (at www.shakira.com) estimate sales as "more than 13 million". Which seems very reasonable.

So we think that 20 million copies is so much of an overestimate it is clearly wrong.

© 2007-10, Steve Hawtin et al. Generated 13 Mar 2010 09:57 GMT. This data may be freely copied provided that the source is acknowledged, a link to the tsort.info site is prominently incorporated and the version number is attached (this data is version 1.9.0031).