Sinatra is a single act who has 17 albums with sales of a million or more.

We're in a pure pop era. And yes, the download and the stream have the focus back onto the individual song. The single is king again.

Albums was currently individual works of art, that's created cultural value for their creators. And that endured listening and time. The Beatles, Dylan, Stones, Pink Floyd, M. Jackson's Thriller is good example.
Streaming sales figures are reliable compare to before. I just readed that Elvis's last studio album, which was released on the eve of his death, has sold 14 million copies. 4 million in USA and rest 10 million outside. Now dare to say that isn't true.

You have a problem listing million selling (single) 78 or 45 or both that were credited before the RIAA was even around. This would include many, including Elvis, but especially Frank Sinatra, & Bing Crosby. Frank's own (1944) 78RPM issue on Columbia-36756 is one of them. Frank would re-record it again for Columbia & Capitol, but for a while, it rivaled Bing Crosby's original 1942 version, released as a 78RPM soundtrack album HOLIDAY INN, as well as a single. This single was later to be part of a 'genre' MERRY CHRISTMAS with WHITE CHRISTMAS 78RPM album collection in 1945. After Bing's 1947 re-recording, it replaced the million selling original as a single, as well as in the MERRY CHRISTMAS album. In 1949, the MERRY CHRISTMAS album was also issued on a 10 inch LP ,later on 45 rpm extended play, as well as a single. Later in 1957, al a 12-inch album, release, it got to no.1 on Billboard's LP chart. It later got RIAA recognition. While no disrespect to Frank, this album, in PHYSICAL SALES, from 1945 to 1957, outsold all other Christmas product. As a re-issue in 2023, it has again charted. As a single, Bing's 1947 re-recording of White Christmas is also (still) popular, with strong claims of over 50 MILLION ,from Bing's sales alone. With no disrespect to Frank, or Elvis or anyone else, THIS record by Bing is one of the biggest sellers ever. While your Bing Crosby listing is still a work in progress, with 22 of his Decca singles were million selling non-RIAA ,the other NON-RIAA million selling singles of Frank on Columbia, need to be INCLUDED for this existing site.

Frank Sinatra SINGLES CSPC RESULTS /Your use of the LP Sinatra's Sinatra re-recorded in stereo REPRISE masters are in error. The rush to re-record classics in stereo, instead of re-processed stereo in the early 1960s, found Sinatra, and other recording artists doing it to get played on FM radio. Not always bad but the originals are better. They SHOULD be the mono CAPITOL albums of:
1957 Witchcraft #10 should be 'All The Way
1957 All The Way # 17 should be 'All The Way'
1953 Young At Heart #18 should be on 'This Is Sinatra'
MORE
1940 I'll Never Smile Again # 14 -the original is NOT on the 1959 CAPITOL No One Cares. IT SHOULD BE 'Tommy Dorsey ALL time Hits, RCA VICTOR-LPT-15 (10 inch) from 1949.
Question /
1947 Have Yourself A Merry Little Christmas # 25- Are you sure? Could it be the 1957 version? /Thank you/

Hi Colin!

These are not errors, the choices were made on purpose. We do not split different recordings of the same song. Just like live versions of a song are combined to the original but the value was created by the original, re-recordings are grouped to the main track as well.

Then of course, I'll Never Smile Again wasn't first released on No One Cares to take one example. But it's the first studio album by Frank Sinatra to contain it, so in a CSPC logic, it's the album which inherit the sales of the track. Obviously this is debatable, but we though it was easier to put it that way.

I'm a little surprised that A Man And His Music (1965) has been considered as a compilation and not as a regular studio or live album. I'm not particularly familiar with Frank Sinatra's catalog, so my question is:

Does this album consist of previously released material or was it counted among the compilations because it contains material from different sessions/live appearances, which, however, had previously been unpublished?

I'm asking because that album went on to win the Grammy Award for Album Of The Year. As long as I know the NARAS doesn't consider compilations for that category, except soundtracks. Generally, the NARAS has a rule that an album must consist of at least 51 percent previously unreleased material in order to be considered for such a category. Maybe that rule didn't exist at the time of A Man And His Music.

In any case, this seems to be the only time in which a compilation album by a single artist/band received that prize.

A monumental catalog that even includes EP’s too, and especially I spotted ‘The World We Knew’ from 1967 that seems to be a British-only release that is given a huge estimate of 200,000!! In 1967, when EP charts were discontinued due to low sales, 30k would be exceptional for the number one and this didn’t even register in the by then restricted Top 10 in the UK. This should be amended.
 
There had also been a fairly popular album in 1967 too with that title (although also technically referred to as ‘Frank Sinatra’). I say popular as it only made No. 24 but lasted 23 weeks in the US to which an incredible 435,000 is given. If it was anywhere near that many at 1967 prices then it would have been another gold RIAA award, which it wasn’t. No. 28 and 5 weeks was its album chart run in Britain which is not worth 80,000. Perhaps these estimates can be explained?
 
Maybe ‘The World We Knew’ (‘Frank Sinatra’) had other EP releases elsewhere internationally, but even so 200k is more than double what it sold, probably more than four times!

Hi Hoboken!

These figures are release to date estimates, you are correct about the fact these releases haven't sold as much upon release. For example, our automated scripts/databases which converts every chart into sales tells us that the Frank Sinatra/ The World We Knew album sold 190,000 units while charting. This does take into account already the increased market during the holiday season. It isn't a good catalog seller nowadays (post-1991 sales are estimated on 10,000 units only), but home to Something Stupid it kept selling well during its first years. Also, as we will see below, charts were deflating Sinatra's success back then.

When a market booms (as LP sales in late 60s / early 70s), recent successes enjoy very solid catalog sales, a bit like how mid-2010s albums kept gaining a lot of streams as more users subscribed to Spotify in the following years. That's why Discogs is so usefull to gauge sales over a longer period than just charts. Here are Discogs owners for similar 1965-1969 albums:

Discogs LP owners | Sales while charting | LP era sales (to mid-80s)
12,670 | 957,000 | 1,300,000 Strangers in The Night (Gold Aug 66)
7,449 | 535,000 | 750,000 That's Life (Gold Feb 67)
4,082 | 190,000 | 420,000 The World We Knew
4,353 | 267,000 | 565,000 Cycles (Gold Dec 69)
6,810 | 227,000 | 780,000 My Way (Gold March 70)

There are 3 things to see there:
1) Billboard charts, in the hyped rock era, deflated an adult-favored artist like Sinatra on their rankings, suggesting below 300,000 sales for albums that went Gold in a year
2) Discogs (as it is the case for all artists) provides us a thiner view of the relative strength of albums, basically US owners multiplied by 100 to 120 gives us LP era sales for every album
3) The World We Knew sold nearly as much as Cycles, and about 60% of That's Life / My Way.

The same process was applied to every format / record / country. That's how we can untape the 'hidden' great sellers like the EP The World We Knew. That one wasn't the British one though, it's sales are mostly from France where it went top 10 of the single chart, almost exclusively on the back of the EP (it accounts for 94.4% of its total EP + single owners on Discogs). Hope it helps!

Fascinating overview and explanation, thanks for taking the time Guillaume.
 
I followed the link you had for ‘The World We Knew’ EP and it went to the UK release, hence the confusion. It should be amended to the French version which is a different track list altogether and then it makes better sense.
 
The Discogs idea is interesting but I don’t quite see how the 94.4% comes about. There are 508 owners of the specific French issue and 612 of all versions (although I don’t know where else it could have been issued so that in itself is a puzzling total). Even Discogs isn’t listing any others it seems, and I might not be using the site correctly, but the two editions shown are the same apart from subtle cover differences and together add to the 612 - 100%. Again, if I’m understanding what Discogs is doing!
 
So, to conclude, you are basically saying this EP sold an estimated 200,000 in France? There is a 0.5 factor alongside the entry that I don’t quite follow either. Is that saying the total must be reduced by half when it comes to the album equivalent sale? Or has that already been applied? I’m sure it is explained somewhere but I can’t find it. Thanks

Actually as they had the same format / title / release year, the UK and French version got combined into the same line in our sheet, so the first link came out, in that case the UK one. If I had noticed back then that both had different tracklists, I would have set a distinctive title (like The World We Knew [EP, FR]) to get two lines instead.

About the 94% comment, indeed 500+ owners for the French EP, compared to 32 to the French single format. Back then, French charts combined singles and EPs, that's why I said the French run was almost entirely due to the EP format, as the single is much weaker as you can see.

Then correct about the 0.5 factor, it does mean that EPs are weighted as 0.5 albums, so the 200k sales of The World We Knew EP are valued as 100k CSPC units in final totals which are based in equivalent album sales.

I do like the Discogs idea and that explains everything neatly as to how Chartmasters works on the admittedly niche area of EP’s. I don’t think you can be expected to breakdown each EP with differing track lists, that would multiply the work so much for little gain in final results. Most will be fine.
 
It is just the link that needs fixing so, if anyone is unlikely enough to wonder off and check these things in the future, they will be taken to the biggest contributing release to a total, especially if it is such an obscure hit such as this case. France is the main place it sold, after all.
 
Linking to the UK issue was more of a problem to explain 200k when EP’s were virtually dead by 1967 there; it made no sense. The fact there were variations is not that important unless they were very much a 50-50 contribution. For instance, it seems Spain had a successful hit with ‘The World We Knew’ and that was probably EP driven too. That release, like the secondary Discogs entry I mentioned for France showing 104 further owners, is important in the way Chartmasters is using data from that site to estimate sales, yet there is no entry for Spain on either single or EP! Do we then conclude it sold nothing in that country?
 
Specifically relying on Discogs, the EP has one French listing owned by 508 people, then 104 own another entry with a cover difference. That is 612. Then nothing for Spain or anywhere else according to the ‘all versions’ link. The single is similarly missing in Spain despite being a hit there too, while in France it has more than the 32 you linked to, it has 140 here:
 
https://www.discogs.com/release/4075176-Frank-Sinatra-The-World-We-Knew-Over-And-Over
 
If the ‘promo’ entry is misleading that is because it is! There also seems to be a jukebox version with 13 owners, but let’s not concern ourselves with that one. So sticking with the French releases, Discogs owners information is suggesting 172 singles against 612 EP’s that is a very different to 94% and nearer to 78%, so not quite so exclusively an EP country in this occasion as you maintained.
 
Ultimately what matters is the accuracy, as best as possible, of the final total that feeds back to the CSPC calculations, and the 0.5 factor is good in that regard. But it remains something of a guess for the 200k, don’t you think? I originally said it was maybe four times too high when thinking the UK and some other lesser releases were being estimated. Now, because it is known to be a different track listing, and even adding in Spain (which I suspect has a different track listing again), the total must be lower, especially as the Discogs percentage is not so high versus the single? I’d say 100k given the size of the two markets is as generous as can be for this EP.
 
This is small fry of course and hopefully you are able to see what I’m driving at with this one example among thousands. But the principle is the same if everything adds up to ruin the final conclusions if it is a repeated error elsewhere. It is so difficult to get right using Discogs alone to get to 200k or whatever the estimate might be, is there nothing else you rely on?

Hi Hokoben!

Yep, understand that about the link, it's just that the script automatically takes the 1st entry, so we have to distinguish them to get the correct link.

Anyway, back to the figures, it's great that you are willing to check the details and get into all the numbers! So to answer the additional points, we do ignore both Jukebox and Promo versions, as these aren't sold to consumers. I know that Discogs is not perfect, and some legitimate releases are flagged as promo, but it's quite rare and has little impact in the big picture.

About The World We Knew, I'm afraid it's trickier than that 🙂 the thing is, Discogs' country reflects the country that produced a version. Often, the version was sold there alone, many times though it was exported to other markets. To identify that, you need to check the Sellers of a release, to see where they are located. For example the single that you mention (the one marked as Promo), is listed as a French release, but sellers reveal it indeed so in Belgium (12 sellers from there, against 4 in France, giving the respective size of the markets, these would be border residents or people who moved since then).

In the same way, we need to look at the French EP's sellers. There we can see that a majority come from France, 66 out of 114, but that leaves a lot of sellers from elsewhere, which is why the EP ends up as a 200k seller while it has indeed not sold that much in France alone (it topped 120k sales there).

While it may look like a 'guess', in reality out of a discography like Sinatra's, there are countless of data points, both from charts, certifications, sales recipes, to accurately gauge the worth of Discogs owners and sellers when dealing with a record for which we do not have as many direct receipts, or even none. BTW, French sales of TWWK are confirmed, no guess there. My own chart convertion tools suggest 135k there, and top-france (which is mostly accurate for singles) put it at #47 for the year with 125k-ish. Discogs' owners/sellers when compared to other singles and albums from Sinatra put it in the same ballpark, so it's a confirmation method more than an estimation method in this case.

For the sake of comprehensiveness, it's worth noting that Discogs' indicators are not as straightforward as what I describe. Discogs' app is used more in some countries (say in the Netherlands), less in others (like Australia). Also, different formats aren't equally collected, EPs survived much less than singles, an equal numbers of owners for 1 EP and 1 single from the same year and the same artist would suggest about two times more sales for the EP for example. So it is by no means easy, and it requires a lot of experience. Luckily we have been digging this matter for many years now, enough to get a decent clarity out of these indicators!

Fascinating explanation of how this aspect of your estimates works regarding Discogs data Guillaume. I think you should consider expanding on more of these Chartmasters estimating methods for people to better understand and, if appropriate, point out what could make things better.
 
The link issue is probably just one bad apple in the hundreds on Sinatra alone, so it is just bad luck that it made no sense. But the promo is a failing at Discogs that, as you say, shows they aren’t perfect and that, from my viewpoint, is the biggest worry with their data. It is all crowd sourced, so very much like Wikipedia it is only as good and as precise as the knowledge of the person making the entry. Many will be very good but equally many other entries will be poor.
 
But most importantly it has been taken over by dealers. Just like businesses have been using Amazon or Ebay these entries are not reliable in telling us where the original record was sold and bought. And that is what your hoping to establish, isn’t it? You have quoted how ‘The World We Knew’ listed mistakenly as a promo single from France has 12 sellers in Belgium. Well, you praised me for going into the details and checking the numbers so I did so again with these 12 Belgians and found all were dealers. I say this because the records for sale are in the thousands, one dealer ‘ararauna’ has 29,766 items listed for sale, while the least number by ‘Ferrard’ is a more reasonable 2,687 but to me that is still semi-professional. A quick look at the variety of records on offer tells us that is likely too.
 
There is no doubt these sellers you are relying on are handling records that had passed through many hands, and certainly would be the case I would think over the 50-plus years for this 1967 Sinatra single pressed in France. So where they are dealing from gives no certainty at all as to the origin of the record all those years ago. You will also see that of the 12 entries three are duplicates, with ‘Vinylsduplateau’ having three listings, while ‘Erikvandevoorde’ and joerids have two apiece (the latter with 21,384 records listed for sale).
 
Do you not agree that it really is not possible to know where these were originally from? They could quite easily have been collections bought from France near to the border as you said, and at any point over the 50 years, and of course dealers also sell among themselves too. There are four ‘shop fronts’ in Germany and one lists 1,198,936 items for sale, and he’ll certainly not have a clue where his Sinatra single came from originally either.
 
So I’d not be inclined to use any seller information about secondhand releases as any sort of guide about possible exports or where they were originally sold and bought. It is poor Discogs data in that regard when you stop to think about it.
 
Hopefully you can see my point. All the same issues exist among the 114 sellers of the EP. All that can be said is it was a French pressing and like so many collectible things they exchange hands and move about over the decades. If 66 are in France then they might be trustworthy but they too are dealers in most cases. I shan’t go into great detail to check completely but the first one, ‘vilzeau’, has a mere 20,968 items for sale. He probably has non-French releases in that number but that doesn’t mean to say they were exported to France either, he just happens to have acquired them in his trading.
 
Can I ask if you are simply taking your 200k estimate and then using the 66 in France and arriving at a rounded up 120k? The sum being 200,000 divided by 114 multiplied by 66 = 115,789. Is that where the 120k comes from now? If so, with all the doubts over the origins of the seller information, you can probably see now how that is not good logic, especially as you haven’t fully explained how the 200k is estimated other than ‘countless data points’.
 
I really think an article similar to the many others you have on the site explaining how a specific case is calculated would be worthwhile doing Guillaume. Then it can be plain to see how data is interpreted.
 
I’m sorry to be so fussy about a trivial estimate in the great scheme of things, but all the small miscalculations add up. The 200k is too high for France and you’ve explained why, but now 120k looks bad too because of the doubtful nature of the seller origin data. You reference Top-Info where ‘The World We Knew’ came in at No. 47 in their rundown of 1967 hits, but how does that equate to 125k, or the 135k you have arrived at using a chart conversion tool?
 
You’ll be busy and maybe not willing to share how these things are done, but I hope you can reply on my concerns. Discogs has some value but it must be harnessed correctly, as I know you’ll agree. The sellers’ information can’t be a confirmation method that adds value unless I am missing something more. If all the data is largely populated by dealers the further back in time we go then Discogs indicators are definitely not straightforward - on that we can agree 🙂

Answering from the forum now that it's available 🙂

You are right about businesses largely using Discogs nowadays, way more than first hand users - wrong about its implications though!

We are talking about second hand records, in an industry where shipping costs are fundamental. These dealers are not buying vinyls from a foreign country to sell it back on Discogs. Instead, people reach their (physical) shop with their pile of old records, and the dealer add the catalog to Discogs. I did it myself with my old CDs when I went leaving in Africa.

I've check tens of thousands of records through the years on Discogs, and sellers happen are incredibly aligned with success in every country. I hate 'believe me' type arguments, but just go through various records and you will see. There are many vinyls with 100+ sellers with virtually all of them from the country of origin. It reflects that when a record was sold only in one market, we still visualize it very well through Discogs' sellers. Instead, when a record was produced in one country and sold largely elsewhere, as with The World We Knew, it gets very clear on Discogs' sellers.

The fact that the same retailer can sell multiple copies of a record only solidy this situation. It means multiple people sold them these records in recent times. Ultimately, and despite low absolute numbers, the sellers really work as a great panel / polling system.

Another important point is that when the data is built, it's never about one single record. For France for example, there are a lot of data available (direct data, like charts and certs), the point is first to estimate albums/singles for which we have enough primary data, then look at their indirect indicators (could be Discogs sellers/owners, YouTube views, Amazon ratings, depending on the country/era) to value how they convert, and then apply it to more modest releases. When you verify a pattern through 10, 20, 30 or more albums, you can be very confident in applying it to the remaining albums. Obviously if the method brings unexpected results (say an album selling 100k+ without charting and no cert), I'll cross-check additional data, search for its release history in the country, other releases from the same label/year, etc. 

As with owners there are some caveats for sellers, namely retailers specialized in imports. These are indeed irrelevant, as they do buy/sell purposedly. I faced multiple of them while working on the Beatles for example, as they are of course a profitable business for these retailers. This tricky is quite easy to identify as 100% of the cases I've met have always been retailers specialized in Japanese imports.

By the way I did write an article, 'Wild guessing is over' IIRC, although that was several years ago and my usage and how Discogs is used in general evolved since then.

As for the more concrete example of TWWK, the reference is top-france.fr, which has been compiled by Fabrice Ferment, known as the French chart expert in France (the consultant for chart-related TV shows, and author of '40 ans de Tubes'). You can see the 1967 ranking here. It shows as an insecure website (it is getting old) but it's ok to browse. Ferment works with labels' receipts, at times they are incomplete, especially for albums, so take numbers as minimums. 

Then the reference to 135k was worked out with my chart position converted into sales system, which considers the market size, how sales are spread through the ranking, the good/bad weeks and seasons. I've built 30+ database tables plus some excel sheets to automatically calculate sales for albums, singles, downloads, or vinyls, for distinct periods / markets. It requires a chart history, known market data, and easy data points (like certs) to build the model. The pages to browse these datasets are not available in the frontend interface of ChartMasters due to legal issues, as well as unclean data (like spelling gaps through time that create several entries for a single record), so it remains for internal use. It's the same algo that I referenced with US album sales, see the screen shot (for old French singles I use a spreadsheet).

The main thing to understand is that this is one big process, with a kind of machine learning methodology. We first dig and write down the safest estimates, then the next safest, etc, progressively we learn and define more and more indicators so once we reach minor sellers we ain't naked anymore!