MFR

There actually is a monthly chart produced for the industry, at least I presume it still exists. These charts cover 4 or 5 sales weeks (Sunday to Saturday) with March, June, September and December being the months that get 5. There can't be any doubt that Moves Like Jagger will top this chart for September. It will also end up higher in the year-end chart than most of this years number ones.

All this depends on a number of factors. Are they including downloads? If so, up to what date? A key point is that sales figures historically were estimates, open to re-interpretation if better information was found. The sample was too small to be certain of total sales while still being big enough to be the accurate weekly guide to relative performance that we know of as the chart.

If they had used 17 as the multiplier in 1992 Stay would have been estimated to have sold nearly 550,000. But I think 15 was not far off reality in 1992, certainly that part of it. OCC had little or nothing to do with the fact that it was 15 at the time as Gallup were calling the shots then. Multipliers are something you need when the chart sample doesn't cover every sector of the market, so basically you have to estimate in that situation. There just wasn't the money available to have as large a sample as today. It's less easy to answer the question "So if it sold x in your sample how many did it sell in the whole country?" than it is to answer the question "So what does your sample suggest sold the most last week, and the second most?" As for singles like I Swear having either 510,0000 or 710,000 (or thereabouts), well it's the result of 2 different methods being applied. The 710,000 is an estimate for the whole market. The 510,000 is an estimate covering only a part of the market (the defined universe, which was the parts of the market that were represented somewhere in the sample). They had this figure at the time as well, but the percentage of the estimated market was too small to quote in Music Week until 1997 when it was up to 99% on average for singles and 95% for albums. Yes, it causes confusion when they quote these figures for 1994-1996 but they are not new, they just didn't get published at the time.

I wonder whether they've used 17 as the multiplier for 1992 and 1993 instead of 15. As they've used DUS for 1994-1996 it would be very generous to mark up some Gallup-era hits. Apart from missing some Westlife hits, as has already been mentioned, there were a couple of NKOTB singles from 1990 that might have been included, one certainly.

I'm not sure that any of these sorts of chart should be taken as gospel. There are too many uncertainties. There were errors in the BMRB charts. Their year-end charts, which have often been used as the basis for longer-period charts, generally only covered the first 49 weeks of the year or so, without subsequent updates. I am reliably informed they didn't even collect the data for the week that no chart was published because of Christmas. More generally, sales occurring in the year after the single was a hit are often not included. OCC only have full data from 1994 and sales for the first 3 years is not in the originally-published format, but at least they are able to add downloads to whatever historic start point they have been given or have accepted.

I sometimes thought the record companies didn't realise, but I think all the majors have applied for singles awards using downloads at some point since 2005, just not very often in some cases.

This topic is currently raising a number of interesting points about BPI awards. BPI certifications for singles are available to all BPI-member companies on the basis of total physically-shipped copies, plus downloaded copies. Prior to about March 2005 only copies actually shipped to the retailers (trade deliveries) were eligible. There is no charge levied by the BPI for certifications. The right to apply for official certifiations to the BPI is part of the membership subscription package. Actual presentation discs do cost, of course. It's not clear why many record companies rarely, if ever, certify singles these days. Maybe someone on here can say how easy or difficult it is for a record company to know how many copies of each of their tracks has been downloaded at any moment. Obviously this information comes through eventually as they need to know what royalties are due as they do for physical copies, but royalties are paid quarterly in arrears so the information from the various sites need not be instant, and perhaps it isn't. Certainly the implication in 2005 was that OCC data would be used when adding downloads to physical sales when calculating whether a single was eligible for certification, and OCC would be likely to charge for that information. Maybe that's not applicable now. Beyond that, there may be a feeling at some record companies that singles are not worth bothering with for certifications. Many albums thse days have a different producer or production team for every track. All the producers are eligible to receive presentation discs for certified albums. Maybe record companies don't want to present a separate disc to the one producer for a single when the album is seen as a project to which everyone has contributed, so everyone gets the same prize in the form of the album certifications. Yes it's a shame that what Music Week printed wasn't reality and was presumably based on translating the known defined-universe sales into the awards that the singles were qualified for. But they can't do that for real because it would make it hard to distinguish the actual certifications with the 'imagined' ones, and would probably kill off the real ones as well if it was a regular feature. Maybe, that's the real problem. It's easy to find out how many copies a single has sold these days. So many companies maybe don't feel the need to spend the money on the discs to tell us what we can already work out if we want to.

After the Steve Miller Band / Deee-Lite controversy the rules were changed (from January 1991 I think) to allow tied positions to occur. However, they were only considered to be tied if the panel sales were the same to one decimal place, so with the Steve Miller Band having 2595.2 panel sales and Deee-Lite having 2594.7 they would not have tied under the revised rules.

As I understand it the BMRB did not even collect sales data for the week that the chart was not published, usually week 51 and the week with the heaviest pre-Christmas sales. Therefore the 1.79m copies said to have been sold pre-downloads either estimate the sales for that week by some other means, or use a record company figure which hopefully tells us what was shipped. It also seems that total over-the-counter sales in 1978-1979 for all the other weeks are less than 1.1m.

In actual fact the BPI does not charge its members for applying for certifications - it's one of the perks of having paid their membership fees. Obviously it does cost a bit to purchase presentation discs, and they may be obliged to purchase these for each certification. Sometimes they present discs anyway without having applied for official certification - go figure! As for The Fame Monster, I would suggest a catch-up is quite likely in the new year.

I may be one of the few people not surprised by Mr Blobby's return to the top, because Take That were already becoming first-week specialists, as everyone was soon to become. When the presenter's tease was "or has Mr Blobby returned to number 1?" I was virtually certain that would be what had happened. And first-week sales not being maintained was a cause of some of the other occurences. Nothing to do with samples or chart compilers. The more records hang around the more they hang around.

This is true. Their monthly charts covered 4 or 5 week periods, as dictated by publishing deadlines, and were most useful when they aligned with other information such as Hit Music's bi-weekly Top 30s.

Not sure Don't Stop Believin' is top of the download list. Rage Against The Machine's Killing In THe Name overtook it. Maybe Journey have gone past again now, but I don't think either is anywhere close to a million. Don't Stop Believin' had sold less than 30,000 before downloads. Rockstar has recently passed 600,000 sales, and appropriately has also been certified platinum by a record company thankfully still keen to recognise such achievements.

Another common misunderstanding about BPI sales awards is that they are based on net shipments (i.e. returns have to be taken into account in determing whether a record qualifies). Returns may become an issue if certification has been delayed long enough for the returns to drop sales below a certification level achieved but not claimed. But this week (May 21st) Katie Melua's new album has been certified silver even though not officially released until May 24th, and obviously there are no returns at this stage. It therefore has achieved pre-release orders of at least 60,000. In the unlikely event that returns ultimately lead to its net shipment being less than 60,000 the silver award will still stand.

It's nothing to do with Music Week. It's up to the record companies to apply for certification to the BPI for these awards, and sometimes they don't bother for a number of reasons. You could always try contacting the record companies concerned to ask them about a specific single or album. They may just have forgotten. They may have presented something to the act without making it official with the BPI. Or they may not be interested.

This is what I have calculated from the available data. 1985 67.8m 1986 60.7m 1987 60.8m 1988 53.5m 1989 58.9m 1990 57.9m 1991 56.6m 1992 48.2m 1993 47.7m 1994 51.4m 1995 65.6m 1996 69.3m Note that the BPI used 15 as the singles multiplier from 1992-1996. OCC inherited that value from Gallup but eventually used 17 for 1994 and 1995, and 16 was used by Hit Music for 1996 (their year-end data for 1995 was also at 16, but subsequently discarded in favour of 17 later in the decade). As I recently posted on Haven, the 1995 figure using a multiplier of 17 gives about 74.3m, which is higher than the 70.7m shipped. The DUS figure for 1996 is 66.2m, by the way. Additionally, it appears that after the event Alan Jones and Hit Music began to use 16 as the multiplier for 1991, although Gallup had used 17 and so had the BPI. I'm not sure what multiplier was used by the BPI for 1985-1987. Alan Jones used 18, but I've been told Gallup's official multiplier was 17 and that Alan was interpreting the data differently to them. Presumably the BPI used Gallup's multiplier, whatever it actually was. All of the above means that sales of individual singles, singles artists and other related data are not necessarily calculated using the same multiplier as the BPI were using for a given year. Indeed most individual titles from 1994-1996 are quoted using OCC multipliers, or at times the DUS.

These do actually exist, the BPI books do contain them, but - 1) You have to calculate them from 2 different tables in 2 different sections. As the figures in each table are rounded you get a significant margin of error. 2) They date from the years when market estimates of retail sales were calculated by multipying panel sales totals by a 'multiplier' to give a best fit figure for the market. The multiplier used by the BPI statisticians is not the same as that used by the chart compilers, or the likes of Alan Jones, in every case (although often they are). If anyone is still interested, and maybe some will not know what I'm on about, I'll try to fish out the figures from my own notes.

I think it's almost completely variable nowadays. The main thing you can bank on is that sales with be high in the run-up to Christmas and more recently just after it. In the 1960s sales were low in the summer and high in the winter, because music was less portable and so people bought more singles when they were likely to be indoors. By the 1980s it was clear that there was a slump in January, which had been a stronger period when the charts were slower and there wasn't a battle for the Christmas number one. In the early 1990s, before the continuous instant number ones era of the later 1990s when individual new releases made a big impact on the overall totals, the pattern was for quiet but building sales in January to early February. There would be a peak around Valentine's day and the school half-term, then quiet again until the Easter school holidays. Although the Easter holidays usually had two weeks the first week was the bigger week most times. After that things were generally fairly quiet until August, with maybe the summer half-term being stronger than the weeks around it . Sales would often build in the later summer (purchases by holidaymakers / overseas student may also have helped). Back to school in September and sales usually dipped a bit but generally higher than in most of the first half of the year (except Feb and Easter). School half-term saw another peak, then often one quiet week and then the build-up to Christmas began. These days it's probably known as the time that X Factor related material hogs the top spot most of the time.

I'm still not sure whether we know for certain that this year has 52 or 53 sales weeks. Historically, it was common for Radio 1 to broadcast a year-chart that was one week short of the full year (it was 2 or 3 weeks short prior to Gallup's tenure as it took longer to compile charts in those days without sophisticated computers). Obviously, radio wants to broadcast these charts in late December or very early January rather than later. Late January would have been the earliest possible time in the 1970s. In Gallup's time the last sales day of the year was the Saturday between December 28th and January 3rd inclusive. OCC moved this to December 27th to January 2nd inclusive as demonstrated by 1998 and 2004 being 53-week years rather than 1997 and 2003. This year-end that Saturday would be January 2nd 2010. Perhaps because it's also the decade-end they may have chosen Saturday December 26th as the last sales day, but that would be setting a precedent. As I posted earlier, in the days when we didn't have weekly updates or pretty much any sales figures, those 51-week charts on Radio 1, used in conjunction with the full-year chart could tell you quite a bit about sales. By the way, I also don't understand how year-end charts could be published on Saturday when a 53-week year would not quite have ended, but maybe that is not what was meant.

Even if sales up to Saturday January 2nd count towards the 2009 and decade charts, it doesn't make this chart wrong. It merely covers a period that is one week shorter, just as any number of YTD lists that have appeared here and elsewhere over the months have not been wrong just because it wasn't the year-end. The alternative to this would be for the broadcast to be in the first week of January when many potential listeners would be back at work / school etc. OK, there would still be ways of listening to it and the chart would still be on here etc. Back in the old days, when access to sales figures was nil or patchy, charts that covered one week less than the full year were very useful because you could compare the broadcast chart and the full year chart and use the differences to judge sales gaps between some of the records.

Your thinking is correct, but OCC are using different criteria to Gallup's. Since Gallup took over the 53-week years have been 1986, 1992, 1998 and 2004. Note using Gallup's methodology 1997 and 2003 would have been 53-week years rather than 1998 and 2004. As this year has the same day/date combinations as 1998, this should be a 53-week year, but we shall see.

The 1996 figures for the Fugees and Spice Girls from Hit Music were calculated using 16 as the multiplier (i.e. panel sales x 16 = market sales). Alan Jones' figures from Music Week are close to using 15 as the multiplier (15 had been used by Hit Music until the final quarter), but not exact. Maybe he used some hybrid of 15 and 16, but the point of change was not the end of the third quarter / start of the fourth quarter. He wasn't using the defined universe sales (DUS) back then, although he has often used them since. In the all-time Top 100 singles from 2002, and referring to sales occurring in 1996 only 2 Become 1, Gangsta's Paradise, Earth Song and Wonderwall - DUS has been used Spaceman by Babylon Zoo - panel sales x 16 has been used Killing Me Softly and Wannabe - panel sales x 15 appears to have been used

That's a fair point, but in the (fairly distant?) past some of those records would have had longer at the top and kept some of the others off.

As a chart fan with experience going back to the 1970s I have to say that I am not keen on a return to one-week number ones and singles that peak in the first chart week. Part of the excitement was in watching them go up, and perhaps making an unexpected upward movement just when you thought they'd peaked. The early 1990s were so much better than the later 1990s in my opinion, with classic climbs to the top by the likes of Rhythm Is A Dancer, The End Of The Road and Think Twice. Multiple runs of instant number ones made the chart dull due to its predictability. It was often obvious what would be number one weeks ahead of the event. At least the late 1990s had How Do I live by Leann Rimes with its bizarre meanderings. Otherwise too many singles got to number one because of their fanbase without having qualities that could attract the casual buyer. Downloading improved the charts no end for a time, but for me it's getting predictable again.

Alan Jones in Music Week 04/07/2009 has the following (sales to 27/06/2009) Earth Song 1,092,596 One Day In Your Life 801,823 Billie Jean 752,766 You Are Not Alone 587,722 These figures are very close to what Simon has estimated. No others above 500,000 but Don't Stop Till You Get Enough is on 494,675 and is expected to reach 500,000 thiw week.