Counterintuitive: How Netflix Letting You Keep Movies Longer Decreases The Number Of DVDs It Needs
from the neat dept
It's always fun to recognize how some initially counterintuitive logic makes sense. A new paper looking at Netflix's (er... Qwikster's) early decision that it would let people keep movies as long as they wanted, discovered that this policy didn't just help Netflix from a marketing perspective, but also meant that Netflix had to buy fewer DVDs than if it had imposed time limits and late fees. This seems counterintuitive. After all, if people can keep DVDs forever, there's greater uncertainty, and Netflix could, theoretically, send out a disc and never see it again. So you might think it would need to stock up in order to account for these huge chunks of time that people might keep DVDs.The problem with that thinking is that it's only looking at the "market" for a single DVD, rather than the larger market of all of Netflix's DVDs, and the kind of demand patterns it would expect. The researchers looked at that, and noted that by letting people keep DVDs longer, it actually slows down the rate at which they take out new DVDs, which decreases the amount of DVDs Netflix needs. In other words, if there's a time limit, people would return DVDs faster, meaning that Netflix would have to send another one out faster. This problem becomes especially noticeable for "hot" movies, like new releases:
To understand how Netflix actually profits from uncertainty, it is helpful to imagine a handful of idealized Netflix customers, as Bassamboo and his colleagues do in their paper. The only thing all the customers have in common is that when a hot new movie comes out, they all want it. If Netflix imposed late fees on its customers, the company could be sure that every customer who wanted a particular new release would return their last-viewed movie, and automatically be sent that new release, within a narrow window of a few days.
Bassamboo and his colleagues discovered that for a hypothetical Netflix-like service with late fees, the company would have to stock as many copies of a new release as it has customers. Given that all those customers would probably only rent it once, the company would be hard-pressed to either remain profitable or offer a compelling subscription fee.
On the other hand, when Netflix allows customers to keep DVDs indefinitely, the point in time at which they return old discs and automatically request a hot new release is staggered. Some customers are going to return a disc the day a new release comes out, while others will take a few days or even weeks to return their last movie and request the new one. The longer customers wait to request that next disc, the more times Netflix can rent out individual copies of its new releases. It is this re-use of discs, or “multiplexing,” that makes the Netflix model possible.
Thank you for reading this Techdirt post. With so many things competing for everyone’s attention these days, we really appreciate you giving us your time. We work hard every day to put quality content out there for our community.
Techdirt is one of the few remaining truly independent media outlets. We do not have a giant corporation behind us, and we rely heavily on our community to support us, in an age when advertisers are increasingly uninterested in sponsoring small, independent sites — especially a site like ours that is unwilling to pull punches in its reporting and analysis.
While other websites have resorted to paywalls, registration requirements, and increasingly annoying/intrusive advertising, we have always kept Techdirt open and available to anyone. But in order to continue doing so, we need your support. We offer a variety of ways for our readers to support us, from direct donations to special subscriptions and cool merchandise — and every little bit helps. Thank you.
–The Techdirt Team
Filed Under: counterintuitive, dvds, operations, supply
Companies: netflix, qwikster
Reader Comments
Subscribe: RSS
View by: Time | Thread
[ link to this | view in thread ]
[ link to this | view in thread ]
Re:
[ link to this | view in thread ]
Re:
[ link to this | view in thread ]
You get a discount if you rent 5 or more pink-box movies. It turns out that renting 5 costs almost the same as renting 3, only a few cents' difference. (In fact 5 might be slightly cheaper, I forget, but it's no more than a few cents more.) Not only that but if you walk up to the counter with 3 or 4 pink-box movies to rent they'll actually tell you that it costs less to rent 5 and encourage you to go find more movies. Every time, without fail.
When I thought about it I figured that it was the idea that movies sitting on the shelves weren't making any money, so they might as well be in customers' hands. And giving that kind of discount and active encouragement generated goodwill and happier customers, thus repeat business.
However, the Netflix model might be part of it, too. Here's a customer with 5 movies to watch. Chance are he's not going to watch them all in one sitting. The more he's got to watch, the longer he'll keep them. And the longer he keeps them, the longer it'll be before he comes back to rent that new release.
Could be a combination. But I'll tell you, it's difficult to walk out of there with three movies when you know you can rent five for the same price. Even harder to walk out with four when five is significantly cheaper!
[ link to this | view in thread ]
[ link to this | view in thread ]
Re:
[ link to this | view in thread ]
Convoluted ?
[ link to this | view in thread ]
Health Club model?
[ link to this | view in thread ]
Re:
[ link to this | view in thread ]
Re: Lies, Damn Lies, Statistics, and Economic Theory
[ link to this | view in thread ]
Re: Re:
the effect does not always appear, and cold water often freezes faster than hot water. The freezing point is governed by impurities in the water that seed ice crystal formation. Impurities such as dust, bacteria, and dissolved salts all have a characteristic nucleation temperature, and when several are present the freezing point is determined by the one with the highest nucleation temperature.
http://www.physorg.com/news188801988.html
[ link to this | view in thread ]
Re: Re:
After what fact?
Something like this?
I'm rather positive that apple will fall to the ground if you let go of it because I have witnessed this effect before.
or something like this?
(DVD qty) = function of (subscribers, rental_duration)
[ link to this | view in thread ]
Re: Health Club model?
This sounds vaguely familiar .... where have I heard this before
hmmmmm
[ link to this | view in thread ]
Re: Re:
[ link to this | view in thread ]
Re: Health Club model?
Also, if this is true, I would think that Netflix' shenanigans would be painfully obvious to the public, because you'd have some people with items in their queue listed as "long wait", while other people would have the same items in their queue listed as "available" (much like how Amazon charges different people different prices - clear your cookies and log in anonymously or log in as someone else, and all of a sudden "your" prices for particular items start changing.
[ link to this | view in thread ]
Re: Re: Re:
In either case, one part of your link confuses me: But from Wikipedia: I'm curious to know how long Lin Edwards at physorg.com thinks a generation is.
[ link to this | view in thread ]
Re: Re: Re:
Instead of having some unhappy customers having to wait for the films they want to see most, what you get is essentially a revolving door of DVDs that everyone gets to see, in time.
[ link to this | view in thread ]
Re: Re: Re:
So, as an example, if the latest release is on the demand list of 25% of your users, and the average full turn around time is 5 days, then you need enough copies in theory for about 5% to make it work out. However, you can play with that number a bit, allowing for the occassional disappointment (not sending the new DVD to all customers on a given day), and profiting from users who cycle the new material back more quickly than average. Based on my experience (I did yield management in another field for years) they probably don't need much more than about enough units for 2.5 - 3% of the users.
The time duration is key, the flexibility in the system is also key, and historical customer demand can tell you a bunch. Taken "pre-orders" and allowing people to put a desired DVD on their wish list ahead of time also gives you a good chance to figure it out. If you go a little further and data mine all the users who are looking for it, you might be able to spot a longer or shorter turn around duration for those customers, and adjust accordingly.
I can't imagine they need a study to figure this out. It makes me think only that these guys have been incredibly lucky in business, not incredibly good.
[ link to this | view in thread ]
Re: Re: Re: Re:
[ link to this | view in thread ]
Re: Re: Re: Re:
Depends really, one human generation or even ten no matter how you measure it, it is a small fraction of time compared to earth, the universe or something else, 50 years is not that much it is 2 generations or 2/3 of one generation, how do we count generations? as they appear or at the end of life of the last remnant of a generation?
On a different note completely because I tend to connect random things together I was impressed by the 3D Slicer, first because you just need to download and it is already pre-compiled with static libraries so it works just by clicking on the binary in Linux and secondly because the model looks truly gorgeous. Then it made me thing about brains in the last segment(part 6) where he creates a model of the brain cavity, which made me think if only I could find a CT-Scan from Mr. Palenti's skull I could measure his brain, you guys wouldn't have a DICOM set from him laying around would you?[/joke]
[ link to this | view in thread ]
Re: Re: Lies, Damn Lies, Statistics, and Economic Theory
1. Don't ever let cash flow be 0.
2.
3.
4.
5.
6.
7.
8.
9. Don't ever let cash flow be 0.
ps: This is a quote from memory from somwhere I don't recall who said it, but it was somebody commenting on MBA's and I believe he was from Harvard or was commenting on Harvard pointing out what one professor said there in a MBA class.
[ link to this | view in thread ]
So What...
[ link to this | view in thread ]
Re: Re: Re: Lies, Damn Lies, Statistics, and Economic Theory
[ link to this | view in thread ]
Re:
Also, 5 movies for 5 days? It will take you longer to watch them all. This seems to imply you would be more likely to return them (all) late.
So the customers may feel like the employees are trying to help them save money, but really the employees are trained to get more late fees out of the customer.
[ link to this | view in thread ]
Re: Re: Re: Re: Lies, Damn Lies, Statistics, and Economic Theory
[ link to this | view in thread ]
Re: Re: Re: Re: Lies, Damn Lies, Statistics, and Economic Theory
.
.
.
10. Don't let the cash flow be zero.
There fixed.
[ link to this | view in thread ]
Re: Re: Health Club model?
Yes it has. First off, it actually happened to me. Second, I read several blogs where others reported the same thing. The suggestion was made to close the account and open a new one with the same list. I did this and at once the titles were all available. But after a while the throttling began anew on the second account. I always mailed my dvds back the same day I received them.
This was years ago. At first Netflix denied they did this. When several blogs published convincing evidence Netflix changed their tune and admitted they throttled but said it was necessary and would continue. Heavy users could sod off.
Based on the length of my list it would take years to see all the dvds I wanted. Since I could download many of them and borrow others from the public library for free I dropped their service. I can appreciate how those who prefer streaming would not care about this situation so maybe for them it's a good deal.
[ link to this | view in thread ]
"No late fee"
[ link to this | view in thread ]
Re: Re:
neutral rather than good or evil, ya know?
[ link to this | view in thread ]