CSPC: Sales & Formulas Fixing Log – Change of digital singles weighting


Professionalizing Physical Singles Estimates (29/04/2017)

To explain this first cross-artists fixing campaign I’ll be giving some insights of our work at Chartmasters.org. How do estimate sales figures of all records?

The first step logically consists in estimating main markets. Certifications and charts are already known, the experience on those markets is also big so there is rarely home for doubts or relevant errors. In a second step, formulas are used to extrapolate Worldwide amounts. Then, collecting non standard raw data enables to validate each estimate, the goal is to identify figures communicated here and there to fix holes as much as possible.

Once all main markets estimates are set and formulas are up and running, the World total is available. It is never manually changed, this is just a result of all available information within’ the Excel sheet. The last step is made of challenging final results with official sales figures of the IFPI or labels reports – real ones, not claimed labels figures that aren’t legitimate in reality. When a gap is noticed between a World result and a valid known figure, we do an analysis to identify the root cause – which raw data or formula created the flaw. After identifying which estimate / formula is wrong we fix it. Then we check if this new information implies additional adjustments on other records for the concerned country. This process enables to destroy myths of massive Brazilian or Asian sales for example.

As mentioned we have enough experience and knowledge to avoid relevant errors in known markets. The % of possible error will always be rather small. The real deal is with formulas. With Hernan, we know by mind how the album market evolved in all World areas through the decades.

Physical singles sales is much more of a black box yet. We both expected the Spice Girls to easily cross the 20 million mark in that format. To be transparent, the first file of estimates resulted in 23 million units sold, while the article has them on 18 million. In fact this is the figure EMI communicated in 2000. The issue is that the main markets – US, UK, Germany and France – already had a combined tally of 16,6 million with only very well tracked figures, leaving a very small room for sales elsewhere.

I then went deep digging into various old documents and sources to identify exact singles markets figures, to understand how it was possible for the girls power group sold so few units in other countries. Results were pretty shocking – while various off the radar countries saw their album market explode during the 90s, the exact opposite happened in the Singles format that started disappearing from mid-80s in a lot of countries. It feels like the US collapse in 2000 arrived early since we stopped buying such products only 5 years later in Europe, but tons of countries had got rid of them way before the US.

Below table displays two curves. The first is the number of million singles sold per year in the US, the UK, France, Germany, Japan and the Netherlands combined. The second one is the number of million singles sold per year in every other market. From 1973 to 2000, main markets remained pretty stable between 300-500 million thresholds. Sales elsewhere have been highly volatile yet and also collapsed way earlier. They dropped as low as 21 million in 1995/1996, exactly when the Spice Girls exploded. This is less than the number of singles that were being sold in France alone, where the group sold 2,6 million units. The 18 million mark was then fairly logical.

Thanks to this new information, the Excel sheet has been industrialized to extrapolate World sales on the back of main markets estimates not with a fixed ratio, but with an adjusted and precise ratio for each and every year.

The Spice Girls article was published in March 2017, after Whitney Houston and before Mariah Carey, so with old and new methods. This is why formulas gave very different results for the same single in spite of focusing on the same estimates within’ the main markets:

Artist – Mariah Carey & Whitney Houston
Record – When You Believe
Format – Physical Single
Old Value – 1,700,000
New Value – 1,320,000
Update Date –  29/04/2017

As you can guess, what is true for this single is also true for all remaining ones from this period. During the next weeks, I’ll be fully re-vamping physical singles sales with adjusted methods for all artists who sold well in this format during the 90s. Due to the low share of this format in overall CSPC totals – no artist has more than 16% of his total coming from physical singles – it will not conclude on a final adjustment of more than 2/3%. Our target is accuracy yet so we will be fixing everything and keep you updated with all new figures coming out!

27 thoughts on “CSPC: Sales & Formulas Fixing Log – Change of digital singles weighting”

  1. I really admire your great work! I can not find a better, more accurate place of data. I was wondering, though. Will you update Linkin Park’s sales once their new album, “One More Light ” gets released? I would love to see their updated numbers since they seem to be having a bit of a comeback on the charts.

  2. Fantastic job on all the work u have done thus far. My question is will u update each artist when they release new downloads/ albums as well as add new artists to the ones already completed?

  3. I love this page and your work!

    But sometimes i think you hate Fred from Mediatraffic hahaha, but is TRUE! His page is very disappointing.

    Please Michael Jackson and Mariah for next!

    And one more… edit Madonna’s charts! Please, please, please!

  4. i’ve always wondered if the method to calculate how many titles each artist sold would also be helpful in clarifying numbers a bit. something like equivalent total tracks sold where

    each album would be for example equivalent to 12 tracks,
    each physical single on average equivalent to 2 tracks
    and each digital single would be equivalent to 1 track.

    but then it would be unfair for artists who release albums comprised of 16 titles instead of 12 and some release just 10 titles per album. so the equations would not be fixed for all artists. but it would give a nice statistic as well.

  5. Hello MJD! Are Britney’s physical single sales going to be updated as well? I’m wondering because she’s sold considerably more digital singles!

  6. Hey MJ Dangerous
    I was wondering if you’d consider a weighting system for features? It seems strange to me that the featured artist gets the same credit as the lead?

    1. Hi Jake!

      May I say without looking too pedantic that CSPC already weights features and in a way much more intelligent than Billboard does ? :p

      I first though really hard about features when defining my concept before realizing the issue was solved in a native way.

      One concrete example, figures in equivalent album sales:
      2010 Eminem – Love the Way You Lie [Recovery] – 4,520,000
      2010 Rihanna – Love the Way You Lie [Orphan] – 1,830,000

      As you can see, Love the Way You Lie brought 4,5m to Eminem while it contributed ‘only’ in 1,8m for Rihanna. This is obviously due to the album Recovery providing the rapper many sales of which none got awarded to Rihanna. This really makes sense – people watching for the featured artist will buy the song rather than the album, just like all those successful Justin Bieber tunes at the moment, so it is logical to provide him those singles sales. If the public is more interested in the lead singer or at least attribute to him the appeal of the tune, they will go after the album. One other striking example is Uptown Funk, a song regarded by 90% of the public as a Bruno Mars song and in fact Mark Ronson album sold close to nothing. Thus, a CSPC tally will be almost 50-50 between both acts which sounds very fair. This is way I say that by its own nature the CSPC logic perfectly balances featurings with a % that is adjusted as per the perception of the general public! I don’t think I can bring anything better to be honest 😉

      1. Thank you so much for answering :D. I see what you mean now.

        I don’t really think that album sales and single sales are that closely related though. Some artists just seen more prone to selling albums whilst other acts can sell singles well instead. Like for example many singers such as pitbull, Jason derulo, flo rida etc move quite a lot of singles and are established hitmakers but are just not album sellers. So features such as 2chainz on talk dirty who contribute very little to the song would still end up getting equivelant sales to Jason just because he’s not a big album seller?

        Also today the albums sales market is getting smaller and the majority of equivelant sales will come from streaming the single so the feature will end up getting a higher proportion of the credit for the songs success compared to if they had released in the 90s. Whereas if let’s say the feature was responsible for 30% of the song and albums success it’d be fairer coz the proportion would stay constant.

        Also it could lead to double counting? Like if artist a and b both release two duet tracks that are both on each of their albums. They’d each get full points for the songs sales and streams and the songs would essentially be counted twice haha. Like I can’t remember the exact figures but supposing that LTWYL generates 12m single sales and 3m album sales for recovery. It’d be worth 4.2m equivelant album sales whereas in this case Eminem would receive 4.2m and Rihanna would receive 1.2m and it’d look as though it was worth 5.4m?

        But you’re right about how track sales can’t just be split. I didn’t think about that. It’s interesting haha. (Btw im not pressuring you to change your formula or anything I’m just genuinely curious what you think about this, your chart knowledge is quite insane and I’d appreciate your input)

        1. Hi again Jake!

          Album and single success are definitely 100% related. The relation ratio will of course differ depending on the credibility / music type of the artist, but at their respective level each artist will sell better / worst depending on the single success. As seen in breakdowns of the likes ABBA or Bee Gees, hit makers did sell albums – not that many studio ones, but their compilations sold a lot thanks to those hit singles. Those compilations sales are 100% “given” to the main artist while the featured one has no boost from that, which seems fine. Then, if we really look at pure hit makers from the digital era (e.g. with cherry picking available, making their album sales a disaster) a la Flo Rida or Pitbull, they do need popular featuring artists to sell well more often than not so we get back to an example close to Mark Ronson / Bruno Mars where it sounds fairly normal to provide as many sales to the featured artist than to the main author.

          You raise a good point about album sales collapsing, but I wouldn’t tell it that way. Album sales aren’t disappearing but instead translating into streaming too. In other words, if the public is interested in more than the song featuring Rihanna or whoever, they will play album tracks as it happens for Drake, Jason Derulo or J Cole. Those album tracks heavy streams are achieved thanks to the single success, this is truly equivalent to former album sales and this is rewarded to the main act only.

          Also, those feats are added only on “Orphan” part. It was an volunteer choice in order to be able to quickly ignore them, knowing too that the main focus would be put on individual albums results.

          Obviously, there must be limits depending on the featured artist contribution. For example, I exclude collective songs like USA For Africa, maybe I’ll count it for Lionel Richie / Michael Jackson but that’s not even sure, I have to make my mind on that. I have to say that 100% of featurings are not named at Spotify and we go mostly by that, voluntarily to exclude feats with minimal and / or non-credited contribution.

          I don’t like the fixed % though as we get into arbitrary waters something I try to avoid as much as possibly on both the methodology and the estimates. Plus, as you note yourself, it would be unfair to some songs, favorable to others. A striking example is Janet Jackson and her contributions on Diamonds (Herb Alpert, 1987) and Scream (Michael, 1995), both US #5 hits. She was not credited on the first one while she was obviously the main reason why the song did so well. In the latter song she had shared credits, the song is among her top sellers in various countries thanks to her brother popularity. If we follow credits, she contributed in 0% and 50% of those songs successes, in a realistic world, she was responsible for maybe 90% and 10% of those songs sales. Thus, a truly difficult situation to accurately represent with formulas. That’s why I prefer simply ignore minimal contributions and then let the album success / single success split do the work between main artist / featured artist.

          Oh, btw, the likes 2 Chainz, Nate Dogg or so would undoubtedly get inflated results but to be honest I doubt we will ever work on this kind of artists 😉

          1. Thank you for this. I really appreciate it. I kinda get what you mean now. You really put a lot of thought into this haha. And you completely solved all of my problems with it XD

            But one last part though, Isn’t there a way for the songs itself not to be overvalued. Like when I said that if LTWYL generates a certain amount of album equivelants and single equivelants. It’ll be overvalued because 100% of it would go to Eminem and let’s say 50% (or whatever amount Rihanna actually contributed based on the ratio would go to her. And then the single would contribute more sales than it actually had. Sorry if I’m not phrasing this correctly.

            Like if Uptown Funk sells 20m SPS singles copies and 100k pure albums. Then Mark Ronson gets 2.1m album sales, Bruno gets 2m album sales. But the songs only worth 2.1m and yet it’s contributing 4.1m sales across both acts. Wouldn’t it be fairer if the ratio was such that considering its a 2/2.1 split they’d get 1.024/1.076 split? As the song shifted 2.1m copies a certain portion is due to Bruno a certain portion is due to Mark, and the song doesn’t end up getting overvalued.

            Of course that would be overly complicated so I can completely understand why you don’t do that. But like what do you think about the principle of it? Thanks for explaining to me how all of the methodology stuff works btw I really do appreciate it. There’s this music forum I visit a lot and like you’re kinda revered there haha.

          2. Hi Jake!

            No worries, I still had your question in mind it is just that for some comments I prefer having a bit more time available to answer 🙂

            To me there is really no duplication since this is an artist point of view, which is the key to the question. On your message, you use the point of view of the song. You are correct that from Uptown Funk point of view, it sold 2,1 million (using your example) rather than 4,1 million. CSPC articles aim to identify the popularity of one artist and some albums instead. From Bruno Mars point of view, he contributed on a 2,1 million selling song, e.g. it added 2,1 million to his discography. I would see no reason to count 1 million for that song on his tally and 2 million for some other of his songs just because of shared credits while in his own songs there is of course also many more people who contribute too but that remain uncredited.

            In a consumers point of view, I prefer seeing it as 20 million persons convinced by one song rather than 10 million persons convinced by Bruno and 10 million by Mark.

            Obviously, on all singles list, I’ll carefully avoid any kind of double counting!

    1. How though? With the ratio of albums to downloads+ringtones changing from 1:10 to 1,5:10, the same amount of downloads equal to higher album sales equivalents! Eg. 1 million downloads used to equal 100,000 albums, now it equals to 150,000. Just look at Britney Spears, who increased by 4 million album sales equivalents and closing in on the 100m mark thanks to this new formula. Nobody’s gonna lose sales. Quite frankly, everyone’s total gonna increase, whether it’s old acts or new acts.

      1. Not sure how, but look at page 9 (George Michael), almost all of his sales have been lowered. Same on page 10.

  7. What’s on page 9 and page 10 isn’t digital song sales though.

    Page 9 is George Michael losing physical single sales. Just like Britney when she went from 96.9M to 95,8M, she lost around 4M physical single sales.

    Page 10 is Coldplay losing album sales. Like raffi said, everyone is going to benefit from this, whether it’s a big increase (Rihanna) or a small increase (Janet).

  8. As much as I like to see some of my favourite artists’ totals increasing from this new update of digital downloads and ringtones weighting, I think this is more of an inflation. Don’t get me wrong, I truly respect MJD’s hard work and dedication for updating his formula and keeping his analysis as relevant as possible, but with this new formula, a lot of new acts or acts with huge download sales are gonna benefit a LOT with this (eg. Lady Gaga, Beyonce, Katy Perry, Rihanna, Bruno Mars etc.) Their albums will have big increases and will match lots of all-time sellers (eg, Teenage Dream, The Fame, 21).

    MJD, I know one of the reasons you changed the digital downloads ratio was due to the inclusion of ringtones, but ringtones were only relevant for a short time (mid-late 2000’s). Perhaps it would be more better to separate downloads and ringtones and use a different ratio for each? Just my opinion though, as separating the 2 formats might be quite difficult.

    1. Hi Raffi!

      In reality it is quite the opposite – I changed this weighting more for downloads than for ringtones. The technical point of view is relevant, but not as much as the functional point of view. As shown by figures, 6 downloads replaced 1 album sales, so the 1:10 ratio was an issue.

      You see it as an inflation because we have been used to see those low figures for many years, it doesn’t make them valid though! In reality, it is the 1:10 weighting which is a true deflation of 00s blockbuster albums.

      For example, without getting into fans wars but using only concrete cases, has Janet been a bigger artist in the music industry Worldwide than Rihanna and Beyoncé? The answer is a clear no. The deflation of digital sales with the 1:10 ratio makes it look like they are lower. If we get into albums-level, let’s compare Janet. (the 1993 album) with Good Girl Gone Bad.

      US Album Sales – 8m vs 3,1m
      WW Album Sales – 12,6m vs 9,1m
      CSPC Total Sales – 19,1m vs 14,4m

      Is that really fair? Both albums had similar hits in the US, both would have got similar album sales too if released at the same time – hit makers are those which saw their pure album sales damaged the most very precisely due to the cherry-picking of songs. Out of the US, GGGB is way, way bigger. It sold more in pure album sales in spite of a lower market and way more singles. In spite of those facts, the final total is much lower than GGGB.

      One may believe “but Rihanna is not an album seller”. This is wrong. A good evidence of that is Maroon 5. Just like her, they have been insane sellers in the downloads fields. Both Songs About Jane and V are about the 10th top seller of their respective year of release in the US (2002/2014). Their sales are not even close though – 5,35m vs 1,17m. The market dropped from $12,6 billion to $7,1 billion only. This shows very well how hit makers were damaged way more than other artists. If Loud outsold easily Hands All Over, both from 2010 and both with huge hit singles, there is no reason to believe Rihanna wouldn’t have been able to sell 7-9 million with GGGB when they topped 5 million with their debut. Or Paula Abdul 7 million. Or Tiffany 4 million. In the same way, it is clear that all 2007-2014 albums by Maroon 5 which were full of massive hits are downgraded with CSPC totals are only 6-8 million, a level ‘flop’ albums from the 90s like Evita were reaching day in day out.

  9. Hello MJD !!
    I have a question for you ! I think that only official videos of a singer or VEVO videos should be counted, because many YouTube users delete their video clips(songs) from their favorite artists and then lose sales / views for some singers or groups.
    I hope you will answer, I am curious about your opinion.

Leave a Reply

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