The Sometimes Fallacy of The Long Tail

There's been a lot of talk (maybe too much talk, to paraphrase Bono) about The Long Tail and how it changes everything about what people consume, how hits are made, what people want to hear, how everything big is small again -- but people have taken that perhaps too far as Lee Gomes contends in a recent blog post about hits. Lee's piece is well thought-out, and I think raises a very valid point that whereas there is value in the Long Tail idea, sometimes people take it too far and that "Hits" still count for a lot. His earlier piece is a more direct critique of The Long Tail and worth reading as well; we covered that piece about the Long Tail a couple weeks back.

Comments

the long tail isn't an amazing concept

[circletimessquare]

most people behave in flocks, a few don't

that's it

it's not like the internet is going to come along and change simple human psychology:

1. the internet is not going to make less people behave in herds

2. the internet isn't going to make more people behave independently

take an old concept, spruce it up with a buzzword, and people think gold has just been discovered. pffft

perhaps the most ironic thing about the idea of the long tail is that the concept itself is now a mass media driven success story with a herd-like following

Re:Is this news?

[2nd Post!]

The interesting part is that the linked article 0.8 percent of NetFlix inventory generates 30% of the rentals.

That means of course that the remaining 99.2% of the inventory generate 70% of the rentals. If they "got rid" of this catalog they would lose a lot of customers.

Of course the data is very coarse, as Spider-Man II happens to fall into the 99.2% category. They need to start analyzing in terms of percentiles; 0.8% generates 30%, 1.5% generates 40%, 8% generates 50%, 17% generates 60%, 42% generates 70%, 78% generates 80%, etc, all the way up to 100% of the rentals.

Re:Is this news?

[TubeSteak]

I don't see how that's "Bad news for hype-driven marketing and economics people"

It's great news for them. It means that if they build up enough hype, whatever they are selling will become part of the 20% of products that generates 80% of sales.

Anyways, the long tail is only relevant to the brick and mortar world if the store owner can afford to keep lots of slow selling product in stock.

Re:Hit's don't go away, of course

[SQLGuru]

And, with accurate forecasts, one's inventory levels are also managed appropriately so that one can carry a very few of the items at the end of the tail such that the next 11% of the profit is gained through only 11% increase in inventory. The incremental cost of that next level of profit is reduced by as much as you gain. So, keeping 100% of the profitable inventory is the way to go.....once inventory is no longer profitable, liquidate it and move on.

Layne

Re:Talk about IRONY

[Snarfangel]

We'll know in a few years. If it still sells a few hundred copies per year a decade from now, in addition to all the other books that would normally go out of print in that time, it may prove one aspect of the thesis. If it's still available on print-per-order, then he would have disproved himself.

Area under the curve matters, not tail length

[Animats]

If anyone talks about the "long tail", ask them if they know how to integrate the area under the curve. The simplest number for evaluation purposes is the value for which half the area under the curve is before that point, and half is after. What's that number for movies? For books? For audio CDs? For iPod downloads?

Netflix says that 30% of rentals are from the top 50 films, so the halfway point is probably below 100.

This is a killer issue for companies that have huge hardware inventories. Consider Digi-Key. They have the broadest inventory of electronic parts in the industry, with over 70,000 parts. Which is a big win for them, because you can usually use them as your only supplier. So there's an Internet-based company that really does profit from the "long tail".

Digi-Key doesn't get much attention as an Internet company, but they're one of the most successful ones. They had online ordering early, and it works really well. Not just the web front end, which looks boring but has what users need, like the ability to search by component parameters. They have a near-complete collection of online data sheets. When a part you've ordered previously is about to be discontinued, you get an e-mail, so you can order a final supply before it goes away. And they have an incredibly effective order fulfillment operation. Orders entered before 7 PM (yes, PM) Central time ship the same day by FedEx. They actually do that, consistently. When you order from DigiKey, you get a confirmation e-mail when the order goes in, and another when it ships, with the FedEx tracking info. The shipping confirmation often comes in within fifteen minutes of placing the order. Now that's operating on Internet time. And that's for orders which might contain twenty different electronics parts in small quantities.

That's a real "long tail" company.

Re:"fear of offending the flock"

[bowmanje]

Does "true independence" even matter? It is not necessary for someone to be "truly independent" in order for his point to be valid. The simple fact that people would consider a purchase that can be made in relative anonymity when they would not otherwise speaks to this.

The long tail and eMusic

[Alioth]

It's funny then how eMusic, despite only exploiting the 'long tail', is the number 2 music store behind iTMS - easily beating all the online music stores selling the popular stuff. Of course, that might have something to do with eMusic being the only music store other than iTMS which sells music that will play on an iPod.

Choice and Overhead

[mugnyte]

I was surprised to see that folks stuck to hits when Rhapsody scanned their usage lists. I think many sites are trying to entice folks to follow the "sounds like" and "genre" trails for new music, but in the end Promotion does indeed play a role.

  Promotion on the internet may best come from "word-of-mouth" sources, where hot links of the week propogate through blogs and link-lists and such. The SNL rap-spoof is just such an example from earlier this year. Every past popular web meme is just such an example. I think podcasts(audio & video) are underrated, since the future of web may be to "leave the chair" and use portables to continue the experience.

  Perhaps in the future a portal will present a Busker-style downloadable playlist of music, completely copyright free "for personal use and sharing" - and then simply ask for donations that go directly to the artist. Popularity and payment mixed into one. For the large portion of music history, this is how such performance careers worked anyway. Then at least a band could make use of the File Sharing networks as a promotion layer, rather than constantly having them viewed as evil.

  Like the Long Tail purports, lowest overhead can make the most profit from the smaller market segments. I cannot see a lower operating margin than online distribution from the artists themselves. The sacrifice is promotion: Labels fight hard for airtime, shelf space, billboard pasteup and DJ chatter. There is a strong argument that "the music world is full of crap and the public wants someone to cull the list for them." But perhaps if these concepts of web meme, genre/artist linking and a modern payment system all converge, things will change.

Re:Is this news?

[Kadin2048]

This is a good point, but I think people are looking at this from the wrong "end" of things.

It's the 'new hits' that are expensive to maintain in a catalog like Netflix's; the 'long tail' is cheap.

Really the economic consideration is "do I have to have Spiderman II in order to stay in business?" And it would appear that the answer, for now, is a resounding "yes."

The reason that the hits are expensive is because each one is like a pig going through a python. For the first week that a big hit movie is out, rentals are going to be through the roof. In order to meet demand and not have dissatisfied customers, you have to have a huge number of discs (inventory) which quickly becomes redundant overhead when everyone moves on to the next hit. Not only do you have to maintain your supply chain for actually getting the movies out and back to customers (Netflix's warehouses, etc.) but you also have to plan on how to acquire and dispose of or store huge numbers of movies that there probably won't ever be peak demand for again.

A large catalog with a small number of discs for each title is easier to maintain, by virtue of predictability.

In short, if you wanted to be a DVD-rental business and you only had enough capital to invest in 100,000 discs for the next year, you wouldn't want to do hits (where your 100,000 discs would get you through one cycle, and then be garbage). You'd want to try and capture the business from the other side of the 80/20 rule, the other 80 percent of the selection that generates 20 percent of the business.

I guess what I'm suggesting -- and granted, I don't have hard evidence for it -- is that it probably takes about equal resources to pursue either the 'front half' of the 80/20 split (the 20 percent of the selection that generates 80 percent of the business) as it does the 'back half.' The hit-based economy seems like it's a gold mine, but you can burn through a lot of resources trying to sate the public's ever-changing demand for the newest thing.