Wednesday, July 25, 2007

The Product Tax

Your feature ideas are great, and your obvious passion is a tribute, I think, to what we've done so far. So I'm going to have another one of those "out on a limb" conversations here about product design. (It's possibly too much for a blog like this, but...) there was something in a few of your postings that comes up often, and it includes the expression that the solution to a problem is "obvious." This has triggered something in me. A number of new community features will roll out next week and it seemed like a good time to bring you deeper into my world.

There is this idea in product design that can be called a "product tax." It has a few components. The taxes are in addition to the more obvious cost of a new feature -- in terms of the work of designers and engineers. This is what most people think about when prioritizing 1/2 stars or movie privacy. I think inexperienced product developers are too highly focused on this as they prioritize what to build. The first small tax on this is the unanticipated amount of work it really takes to build and support the feature. Take our Movie Privacy feature.

As I said back when it rolled out, the creation was opportunistic: we had built it a year ago and it hadn't been released, it was not particularly "designed" - but it did solve a need and I felt it would be important to Friends users. And heck, Blockbuster even had it. Alright, so it wouldn't take too much time to finish it up and get it on the site, as long as I didn't spend a lot of time making it pretty and efficient. It just "is."

But once it was connected up to other pages, unexpected questions arose: if a movie is hidden, what happens if you review it? Does that "unhide it" or does the review show anyway. What about Watch Now movies? Are those hidden too? Someone discovered that hidden movies DO show up in the RSS Feeds. Should we hide the movies there? Each place where movies need to be hidden requires a little more work to implement. And in some cases, say RSS Feeds, there are engineering reasons why it's harder to hide something than in your Friends' sliders. Now I'm going to make up some numbers to illustrate this, but they are just made up: let's say 25% of our members use Friends. And 10% use Watch Now. It's possible that only 1% of Watch Now users are also Friends users. And not everyone in that group wants or needs privacy-- maybe only 10% of them. So to add this simple Movie Privacy page, and make it work "right" we have to think about and engineer this hiding for Watch Now even though it's highly unnecessary: 25% of our Friends users need Movie Privacy, but only a fraction of a % need Privacy applied in the Watch Now area. (And for the record, Movie Privacy DOES work on Watch Now titles). The point is that it was easy to create the page, but finding all the strings leading away is harder and creates a tax on releasing the feature. We tend to make something work only when a reasonably large percentage of people want and need the feature, or if common sense dictates we do it anyway, expecting more people to be using it later on. But as obvious as the feature is, one needs to be careful following all these strings and considering that additional work in the feature. But as i said, this is only the "small tax" here. I think product designers are well aware of this tax.

The larger tax is the one frequently missed. Once the new feature is created, it adds a lot of software to an already large pile of software that is our website. And now whenever someone working on any other part of the site has the idea for some cool new feature -- one that might be VERY important to lots of people, and VERY cool, the building of it must take into consideration all the other things it touches. As Steve goes and builds Watch Now features, he is belabored with the strings of Movie Privacy. If I want to add something about notes or reviews, I can't just build it quickly and get it out, i have to consider all the permutations that may involve privacy and make sure each is logical. And the more of these things you add, the more difficult and slow it is to do other features.

This is the hidden tax. It's starts small, but expands rapidly. Products (and websites) often get slower and slower to release new features because they spend an increasingly large percentage of their time just dealing with the baggage of the old features. If those old features are great, that's just business. But if those old features don't add much value, or only add value to the 1/2% of people who write reviews, use RSS feeds AND use Friends, then one must consider if the cost is worth it.

The discipline, therefore, is to generate new features, make them good for a lot of people, and keep an eye on how much value they really add. And after some period of time (it could be half a year, or a few years) if they aren't being used by many people, kill them. Otherwise they stack up in an endless pile that must be weeded through by customers and engineers alike.

And there is one more factor to consider: in many cases, adding more features diminishes the usability of the existing features. People only have so much attention (particularly for a website). If I have one button, you can click on the button. But if I add a second button, we have seen over and over again, that often people don't click on either button. There is confusion. There is choice where there was none. Overall user experience can get worse when you add features, and the exact opposite of what you want to happen occurs. We think we've added something people want, but fewer people see the features in total. Weird, but true. Less is more.

You are sincere when you suggest a new obvious feature, and it may even be clear to us the feature is cool. We consider the costs and benefits and then build them. But we also think of the tax, and need to think of the bigger picture, of a site 3 years from now that does all kinds of great things, things we KNOW many people (most people) want. If this cool little thing now is going to impact that supercool big thing then, we have to decide whether this thing is worth it. These are HARD decisions, there is no right answer in many cases.

As I mentioned, my brother is a Hollywood screenwriter, and I was a movie and TV editor in a former life. Writers have this notion of "killing your young" -- a rather crass way to describe the painful process of writing and editing out stuff that you like. For me as an editor, the way I learned it was like this: editing a movie is not about taking out the bad stuff. editing is about taking out the good stuff to make the remaining stuff better. You don't usually see all the stuff removed to make your favorite movies so good. But trust me, some of it is great, and you'd think they were idiots for cutting those scenes or lines. But the result is the thing.

Please keep your suggestions and debate flowing. We all read this stuff and it's good to hear. Many of your ideas are being implemented while you read this. So thank you.

Anyway, welcome to the inner conversation.


  1. Sometimes you just can't catch all the pitfalls either. Good explanation.

  2. Excellent idea to slightly temper unrealistically high expectations at this point, and to explain so succinctly why we might not all be getting that new toy we so desperately wanted.

  3. One question. Who is your brother?

  4. Love you guys and your metaphysical musings! Oh, you meant Michael's.

    Very instructive inner dialoguing, btw - because without understanding the process it seems like it would just be a matter of adding a few lines of code in between snacks.

  5. Thanks for posting that explanation. It's a good reminder that things are a lot more complicated than they appear ... and also testament to the great job you're all doing to make it look so easy with the results you provide.

  6. Hey, this weeks poll is getting cut in half on my screen. Is that happening to everyone else? It is too wide for the sidebar (A common problem on last Weeks poll is the right size, but this weeks is getting cut off. Can't read the questions and answers.

    OK, that's all.

  7. michael, from netflixJuly 25, 2007 at 12:23 PM

    Fixed the weirdness in the poll. should be okay now.
    (my brother is Danny Rubin, who wrote "Groundhog Day")

  8. I'm now having the same problem (Mac/Firefox) with the poll, although it was OK up until a short while ago. And yes, when all WE have to do is click on a tab, it's very easy to forget that this relay switch is triggering an enormously complex chain reaction throughout the site. Appreciate the non patronizing heads up on that.

  9. PS. Sorry, our postings crossed. Poll OK again.

  10. Thank you so much Michael and the Netflix team for trying to satisfy even our most neurotic desires. After reading your post, I guess we all felt a little guilty for demanding so much from you, hence the lack of comments here. As you mentioned, it is very probable that the friends users, movie raters, review writers, post readers, new features requesters, can't live-without- movies kind of people are just a minority. But at the same time we understand and live with you every change, appreciate every effort, and are the most unlikely to cancel or change companies. Netflix is a service that is truly alive, constanly changing, constantly improving. Thanks so much for being as passionate about pleasing us, as we are about our movies. Thumbs up to the Netflix team!

  11. Very nice explanation of a hugely complex process.

    Thanks for taking the time to share that with us, Michael.

  12. Thanks for your thoughts. As someone designing my own social website (, it's helpful to be reminded that even two buttons can be one too many.

    The work you guys continue to do on this site is a huge inspiration to me and my design team. Netflix is a work of art. Keep it up!

  13. My experience the last 18 months with Netflix is mixed. I appreciate the constant work to add new technologies (drag-and-drop in the queue) and features (Sim%), but I've been frustrated with, in some cases, poor execution -- particularly in the area of interaction design.

    Please do not consider this post as a personal attack. I am a designer. You've expressed an opinion about something I do every day. As a user of Netflix, I'd like to respond from my professional standpoint. Thank you for providing a forum for me to do so. ;-)

    I'll start with a short analysis of the example you used to illustrate your point about "taxes". If I read you right, are you saying that people on your team are inexperienced product developers, which is why they (you?) failed to fully think through the logical implications of the privacy feature before rolling it out.

    The very first thing a designer would want to consider with a privacy feature is that the implicit promise of privacy between the application and the user is a critical component of brand management: Can the user trust Netflix? A wrong move there is the biggest tax of all.

    For that reason alone, the attitude going in should never have been "It just 'is'.", but should have been: "This thing has to be bulletproof." The questions that you indicate arose after the roll-out should all have been considered well ahead of time.

    I also have to take issue with the notion that a development team's job is to roll out features, give a couple of months or years to see how they do, and then pull them if they aren't serving enough value. With the active user base of Netflix, you should have plenty of user input to give you an idea of what features are needed so that you never find yourself needing to pull a feature because it's not being used. That's not to say you won't have a vision for something the user base never even considered, which could be a stroke of pure genius and worth taking a risk. But those kinds of things are big leaps. The daily bread of your development should be focused on making what you already have the best it can be.

    When you say that accumulation of features over time can make an application "slow", I am I correct you mean too many options bog down the user experience? If so, I think that's true, but it's the job of usability engineers interaction designers, and information architects to modify the site labeling, navigation, and structure so that features are intelligently layered such that the most critical features are close to the surface, and ancillary features (options really) are available where needed.

    Much of your post seems focused on the assumption that the input you're dealing with from we users is a cry for more features. What I see is more a request that existing features be improved (options or better work flow design).

    Perhaps the inherent "lesson" in your post is that the true discipline is not to manage the process of introducing new features, but to focus on making the ones you already have work the way users would like.

    For the record, I think the design/development team at Netflix has made great improvements over the last few years. However, I do think you could spend the next few development cycles tightening up interaction and usability.

    Why, for example, to I have to pretend I am going to review a movie, just to find the link to "Visit your reviewer page"? Why isn't there a link to that work flow on the "My Account" page? Everything about "me" should be accessible from that page. And when I get to the "Your Reviews & Lists" page, why don't you include the "X out of Y people found this review helpful." information right with the review? Even more, why spit out the entire contents of the review right there on the page using valuable real estate to present a lot of text that I am intimately familiar with since I wrote it? Why not just present enough text (three or four lines) to utilize the horizontal space required by the movie image icon, and have an ajax powered "more" link if I'd like to see the entire text. Doing so would permit display of many more reviews right there on a single page. Why not have a link to show me my review in the context of overall reviews for movie? This is pretty basic stuff that any experienced designer should have been able to see at first glance.

    Yes, I could go on, but you get the point. Sticking with my example (I'd love to move on to "search"), I think the idea to give me a way to process "my" reviews is great, but the execution side seems like the guy with the vision dumped it on a developer to flesh out and never looked at it again. And of course, the result is users like me expressing frustration and asking for changes.

    So, perhaps another tax to add to your list is the tax for not taking the time to do it right in the first place.

  14. This is obviously for Michael to answer if he wants to, but I'd just like to say in Netflix' defense that we can have no idea what suggestions have been pouring in over the years through the 'Contact Us' channel, and to what extent design decisions were predicated on those suggestions. Plus, if 10 people were to look at each one of the things you consider to be obvious shortcomings of the site you're likely to get 10 different levels of agreement or disagreement; one only needs to read 10 random reviews of a movie to see that very little is ever 'obvious' to others in the way it is to ourselves. And Netflix has some 6 million potential reviewers to keep happy - not an enviable prospect.

  15. This comment has been removed by the author.

  16. We appreciate all your work. I am in the 1/2% who would use the Instant Watching and the privacy feature now so I hope someday it does come to be. but for now, keep up the good work.

  17. One question. Why bother to put piracy restrictions in your TOS, if you dont intend to do anything to someone who pirates nearly 400 movies? If every NETFLIX user was allowed to do that.........what's the point of renting from NETFLIX at all? We can just borrow pirated copies from the thieves who make them, right? Just a thought.

  18. Interesting post here… and btw I really enjoyed that TED video in an earlier post. That was a lot of fun.

    I’m so relieved you guys are trying to keep this thing from running amok. For all the things we want or think we want, that is what we will always want… no mok.

    That is why I think this back and forth is so valuable. To help us all understand what is really wanted/needed and what paths to abandon.

    I hope we can continue to help you evaluate what features are really of value to us, and as enthusiastic as we are with our ideas, I expect that you will continue to evaluate them and test them in some way against the broader netflix population of subscribers to avert unforeseen consequences or damage. As long as the level of functionality we have now is maintained, I can easily wait weeks, months, years for new features. Especially if I know they are in development or under continuing evaluation and will be rock-solid upon implementation.

    I also suspect, not being an expert ofc, that this tax situation, is why the bulk of the ‘community’ stuff should be in effect separate from the rest of the service so there isn’t a lot of complicated cross-effect.

    You’re right ofc… Sometimes less is more and more is just useless.

    From your communications here I feel pretty confident you’re trying to keep things working well as development progresses.

    Oh… and what you said about movie editing… “You don't usually see all the stuff removed to make your favorite movies so good. But trust me, some of it is great, and you'd think they were idiots for cutting those scenes or lines.”

    Your damn right! Remember the deleted scene from "We Were Soldiers" where the Sergeant put on his medals for his new Lieutenant? That was freakin’ brilliance and they were crazy to cut it. I LOVED that whole scene. It’s more memorable than the Movie and practically no one has seen it. Araghh!


    Yeah… Good Job Here! Keep it up.

    I have full faith and trust that we can keep things progressing smoothly together.


    PS: Yay “Groundhog Day!” and this is your SECOND career? damn…

    (you didn’t cut that scene from “We Were Soldiers” did you?)

    OH… and I’d like the review text on my review/edit pages collapsed too, so I can find the review I want easier and then expand it to view and edit. Here comes the Tax-Man… ;)