Re. the update: Thanks for including that. It's good to see that there is a path within the legal system for challenging such outrageous legal bullying. (Knowing this provides some balance to the picture that sometimes emerges here.)

----- the company still uses faxes from small record stores to receive information when POS systems are not available -----

Um, which other course of action would you suggest Oricon take? They could either

1) Install POS systems for the small stores and incur the expenses themselves, or
2) Exclude the small stores from their sampling

2) would just lead to a biased sample. Are we supposed to believe Soundscan Japan generously chooses 1) in pursuit of more accurate results?

----- They also count CDs sold at live performance venues and include internet sales. -----

As they should, especially regarding internet sales. Soundscan will eventually start to count venue sales too, if they want to provide more meaningful results. Given that no one has figured out a way to track venue sales accurately, Oricon's methodology is likely flawed, but they're the more progressive company of the two on this count.

----- Nikkei actually provides proof to the claims in the original, controversial Cyzo article that Oricon gives "VIP treatment" to Johnny's Jimusho. -----

Huh? That's proof that Oricon and Soundscan provide different results. There's no more proof of Oricon's bias for the idols than of Soundscan's bias for the rock and indie acts.

----- The gaps may be simple sampling issues, but Oricon's insistence on using non-POS data makes it much more open to suspicion. -----

Nice framing. Oricon doesn't "insist on" using non-POS data; it uses that only when POS data is unavaiable. But I suppose you can say Oricon "insists on" including small stores that can't afford to install POS systems. And I can say Soundscan "insists on" excluding these stores and having a biased sample.

The question is whether Oricon is more "accurate" in reflecting the market by including more sources, but less accurate in the fact that their non-POS data can all be more or less manipulated. If they would just explain where the numbers came from for each act (how much in POS sales, how much in live sales, how much in local shops, how much from the internet), there would be some transparency in the process.

Oricon is supposed to be getting the same POS data as Soundscan, no? Then I find it weird that you see so much underreporting of a certain kind of act that does well in the chain stores. We have no idea how Oricon is weighing its data when it samples for national sales.

"That's proof that Oricon and Soundscan provide different results."

Yes, I agree that the "proof" depends on how deeply you read into. Maybe it doesn't prove a VIP relationship, but it at least looks like you can ask the original question. What situation would provide Oricon with 20,000 more sales of a CD than Soundscan? (Especially when you look at data that's before the age of internet shopping.)
I don't buy the "live sales" theory, because it's not like Johnny's has a concert every week. There is something systemic about the way Oricon oversamples a certain kind of idol music and undersamples a certain kind of indie rock. If Oricon is perfectly accurate, this means that Soundscan is way oversampling all rock bands and undersampling all idol music. But when you start to look at other charts (like Amazon and Tsutaya), you get a sense that Oricon is the only one to oversample these kinds of artists.

Does Oricon consistently oversample the entire genre of Idol Music, or they they show a preference for specific publishing and/or management companies?

I can put up some numbers later, but I see very specific preference for Morning Musume's production company Up Front Agency and also Johnny's. Shibasaki Ko, for example, is an "idol" of sorts, but has almost the same numbers on both charts. With Up Front and Johnny"s, you are looking at a 20-30% increase on Oricon, not a 5% or -5%.

I have my own evidence to support your decline and rigging theories. http://akabe.livejournal.com/24584.html

I conducted the same experiment in winter 99 - 00 and the results were pretty much the same, i don't remember the exact score.

I agree with you about the lack of transparency from Oricon.

I actually agree with much of your underlying message, but the way you've framed the issue rubs me the wrong way. Transparency and accuracy are two different issues, albeit related. Of course Oricon's numbers can't be all that accurate given what we know, and neither can Soundscan's. The real "accuracy" problem here is the POS system's penetration rate, not Oricon.

----- I don't buy the "live sales" theory, because it's not like Johnny's has a concert every week. ----- Also, wouldn't most Johnny's Jimusho event-goers already own the CDs of the artist? -----

At the moment Johnny's have nine active debuted groups. The total number of concerts they hold each year range in the hundreds.

Moreover, Johnny's have this charming practice where, for a group of X members, they may sell up to X+2 editions of a CD with different covers (first press, regular group, individual member). Many fans buy multiple copies, which mostly helps to build first-week traffic. But also, if you already own one copy going into the concert, you may come out with three or four, depending on which members struck your fancy.

"The real "accuracy" problem here is the POS system's penetration rate, not Oricon."

Yes, but add to that the long-standing grumbles about being able to buy chart position through placing advertisements in Oricon. Billboard had the same complaints. Whether these are true or not, Soundscan being based on all POS makes this kind of tampering theoretically impossible (besides the fact that payment for a "Soundscan #1" is a waste of your money.) These rumors do prove that Oricon is up to no-good, but Oricon's methodology does indeed benefit large idol groups from big agencies who most need the "#1" type of badge for their marketing strategy. Kinki Kids at this point cannot afford to have a "#2" first-week single. This doesn't necessarily mean that Oricon is tweaking the charts for the Kinki Kids, but with their methodology, it's hard to disprove those allegations as well. It doesn't help that Soundscan shows a 20,000 or copy gap for these certain classes of artists where other artists' gaps are much less substantial.

"The total number of concerts they hold each year range in the hundreds."

Wouldn't the band in question, however, be too busy with promotional activity for the single to appear in a concert?

Of course if Oricon disclosed their statistical methodology, we wouldnt have to speculate. To further muddy the waters, record companies have different ways of counting "sales" (in the US anyways, I have no doubt that the locals have their own interesting methods). Without all the cards on the table, POS looks cleanest to me.

I don't see why Oricon should have to refute the allegations. Until the POS system reaches similar penetration rate in Japan as it does in America, there is no way to show that Oricon's results are less accurate than competitors'. All people can do (and in fact have been doing for long) is raise doubt. For what purpose? They can't or won't prove they're more accurate; you can't prove they're less accurate. At the moment this debate is about as substantial as "you suck/you suck more."

------ Without all the cards on the table, POS looks cleanest to me. -----

No one said otherwise. Again, Oricon uses POS when it's available. The "accuracy" issue at hand is extrapolation due to less than ideal penetration.

Billboard's numbers are accurate because Soundscan America merely adds up all the POS data; it doesn't extrapolate to cover record sales outside the system. That's not the case in Japan.

----- Wouldn't the band in question, however, be too busy with promotional activity for the single to appear in a concert? -----

It’s not difficult for a group to promote a single on TV during weekdays and do four concerts over the weekend (Johnny’s acts can do two shows per day, sometimes even three). TV promo takes up a few days over 2-3 weeks, but tours may last a month or two.

passerby,

its not a question of refuting alegations, its a question of supporting their claims.

Again, I think it's suspicious that they are both sampling based on the probably the exact same POS data, but you can get 30% to sometimes 100% differences in sales. And this has been true for a long time - not just with the rise in internet sales, etc.

"It’s not difficult for a group to promote a single on TV during weekdays and do four concerts over the weekend"

The could, but are they? We can go back and look for those two weeks whether the acts had live gigs. Also, how does Oricon count the number of CDs sold at a live gig? I'm guessing that it doesn't, and it's the "seller's responsibility" to accurately provide the number of CDs sold.

----- its not a question of refuting alegations, its a question of supporting their claims -----

I was replying to marxy's "...it's hard to disprove those allegations as well." Detractors started the whole hoopla with their allegations.

----- based on the probably the exact same POS data -----

Can you find anything on that? Oricon publishes a partial list of participating stores (http://www.oricon.co.jp/rank/index2.html). We can compare if you can find similar info on Soundscan.

I wouldn't hold my breath though. Oricon taints its sample this way. If Oricon and Soundscan use different stores, this may explain the discrepancy between their numbers. Idol fans may pick Oricon's stores to ensure their purchases count - disproportionately. But that's only speculation at this point.

----- We can go back and look for those two weeks whether the acts had live gigs -----

A group usually releases 2-4 CDs a year but does only one tour. The promo and tour periods overlap at most once. I can give you examples either way. But you're the one insinuating a shaddy Oricon-Johnny's link; shouldn't you research them better yourself?

As for venue sale, I'm guessing self-report too. We assume it because we know no better alternative. Same goes for non-POS stores which supposedly "fax" their numbers in. So we're back to the stalemate. Either ignore this large chunk of sales, or attempt to measure them based on trust (I know, lulz).

[[Billboard's numbers are accurate because Soundscan America merely adds up all the POS data; it doesn't extrapolate to cover record sales outside the system. That's not the case in Japan.]]

I thought Soundscan America DID do some of this though. Years ago I was told that certain retail outlets in more rural locations are statistically weighted so that one unit sale rung-up will count as three or five or more...? Maybe to account for mail order perhaps? Not really sure. Heard this at a Distro. Co. back in the states that handled mid-sized & indies. There were stories of people knowing just which store in Kansas was the one to buy five copies of the new Tahiti 80 or whatever at.

"Idol fans may pick Oricon's stores to ensure their purchases count - disproportionately. "

More like idol companies. And yes, this used to be standard practice back in the Billboard pre-Soundscan days. Companies would figure out which stores were being sampled and buy there.

http://www.geocities.jp/zuckey_tb/Photo_TKO.2006.htm

I found some TOKIO Tour info. They would have played three gigs before the Oricon reports went in - in Ishikawa, Niigata, and Iwate. Not exactly huge markets. Could these be responsible for the 20,000 CD difference? Seems possible, although again, how are these live venue CD sales recorded and verified? Also, 20,000! is a lot of CDs for three venues. The Niigata place can only hold 1730 people.

I can find no indication that Endlicheri did any gigs that week. They seemed to have gone on tour after the single was released.