A Revolutionary Take on ROI, Business Value, and User Experience | Main | My legacy.

July 29, 2004


I'm working on a project where I get to go into people's homes and watch them attempt to set up an internet-enabled device (excuse the vagueness). And, without fail, they cannot. What's interesting to me is how they fail -- each time it's different. Though often in the same part of the process, the detail that causes them to go astray varies -- mistyped email addresses, password confusion, network set up, clicking the wrong link and getting lost, etc. etc. And I'm sure that with each subsequent observation, we'll observe new hitches.

Over 15 years ago, The Donster wrote The Design of Everyday Things, where he wrote about the difficulties we have with fairly simple products -- door handles, stoves, etc. The most complicated single item was an office telephone, with it's various modes that you access through key presses.

Well, of course, none of that has anything on the bizarrely difficult connected world we currently live in. Computers plugged into printers, into cameras, into networks. TVs plugged into DVDs, DVRs, stereos. And the techno-types want more more more. Refrigerators with IP addresses. Smart medicine cabinets. Thermostats that require "programming." House alarm systems that detect individuals distinct biopatterns. Hell, my cell phone doesn't want to make a call -- it seems much more geared to "opening application" now. Look -- I just want to talk, I don't want to *use an operating system*.

Objects aren't simple any more. They don't just turn or push. They behave. And these behaviors are often played out over many steps, in particular orders. And each step is an opportunity for failure. Through the work that my colleagues conducted on business value and user experience, I learned the six sigma concept of "rolled throughput yield."
the probability of being able to pass a unit of product or service through the entire process defect-free.
Basically, this means that the more steps you take, the likelier failure is. Even each step has a high probability of success, when you add them up, the likelihood someone can get through all of it becomes startlingly low.

And that's the big problem I'm seeing. Too many steps. Too many opportunities for something to go wrong. For people to enter a typo. For people to not see and instruction. For people to click the wrong link. I'm not blaming people for this -- they're just doing what they do.

My client assumed they were doing the world a service, allowing you to connect this device directly to the internet -- without a computer. It will make everything easier to have it all in one place. It will allow you access to things you couldn't do before.

But connecting to the internet is not easy. It requires modems, routers, and setting up. To add devices requires cables, or wireless-savvy. People have already gone through a fair amount of hell just getting their current system up and running. Adding to this often just upsets them. People are happy to just have something that works. When you suggest to them ways they could optimize, they'll turn a deaf ear -- change brings uncertainty, change could mean that the thing just doesn't work any more.

In the case of our client, complexity has been bred by a marketing group that WANTS MORE FEATURES (because they sell!) and an engineering group that adds these features onto the existing product like rooms in the Winchester Mystery House -- without rhyme nor reason, with no overriding logic, with no real principles to guide them.

Anyway, the nub of this wee rant is, of course, simplify. It feels like we're reaching a breaking point with new technologies -- if we thought the blinking "12:00" on the VCR was sad, what terrors can we expect? Think about the lesson of rolled throughput yield -- how can we minimize the steps involved? How can we enable people to plug something in and *just have it work*? How can we do a little more work on the design and engineering, so the customer has to do none on their own?

Posted by peterme at July 29, 2004 11:12 PM



You might want to turn to the original WebTV for some inspiration. When I first saw one, I was stunned. You plug it into the wall: a phone cord and an electrical cord. And you turn it on.

It dials an 800 number with its modem, which figures out where it is and ships it a list of local numbers. Then it calls the local access number, handshakes, and--

--goes. Nothing to it. It just happens. (Even AOL never made it that easy.)

Incidently, I saw this at a talk by someone at Berkeley SIMS. But the setup wasn't part of the demo. The group had so internalized the setup as easy that they really wanted to show off some things later in the process, and so they did that before class. I just happened to be early.

Ever since then, that's been my gold standard for a usable setup for a technology. Of course, I want the thing to tell me *why* it failed if it goes down ("ack! no phone cord!" "4039156 isn't a local number!")--but the default case should Just Work.


Posted by: Danyel Fisher at July 30, 2004 02:24 AM

There's a delicate balance here. Don't err in the other direction, where the thing fails to work and you have no way to know why.

See Eric Raymond's diatribe against the "easy" configuration of CUPS - http://www.catb.org/~esr/writings/cups-horror.html

As he points out - the problem isn't that the grandma's can't do it - the problem is that the uber-geeks can't do it!

There's another issue here which you haven't really touched on. There's a divide growing between types of consumers:

  1. those that want the product to just work, in and of itself.
  2. those that are willing to accept the reality that the product is part of a network of services which may be interrelated in one or more ways

Type 1 consumers will get confused if the product doesn't work out of the box (or blame problems with a service on which your device relies on your product directly), and they're the reason why your device connects to the internet directly. This answer will frustrate type 2 consumers, who will take the position that the product is broken unless it can integrate with their existing network of things.

I wrote a short essay a while ago in which I babbled about the fact that we seem to be leaving the age in which consumer devices are standalone products - http://www.someoftheanswers.com/dynamic/qa.html?id=16

Posted by: Adam Fields at July 30, 2004 03:07 PM

Love it! Finally some intelligent debate about a key issue for the future of technology.

My own rule for this to help explain to clients why their digital efforts often fail is that:

"The overall use of a system is directly proportional to the ease-of-use of that system."

The 'system' being defined as the business' complete service delivery and not just one element of it.

Also, as I discuss in one of my recent posts (http://alteraxion.typepad.com/movies/2004/07/confronting_cha_2.html), there is a noticeable increase in the number of 'unfinished' products and services that are making it to market of late. These are putting a significant strain on consumers who must walk the last mile to figure out how the whole thing works.

Imagine if you were to buy a toaster but had to fit the heating element by hand... how many of these toasters would they sell!? What would be the out-of-the-box consumer experience?

Posted by: Andrew Carton at July 31, 2004 04:21 AM

"The overall use of a system is directly proportional to the ease-of-use of that system."

This is technically true, but not the whole story. My version would be "The overall use of a system is a direct equation between three factors - F * V * E"

F = Functionality of the system
V = Visibility of the system (can the user see that there's worthwhile functionality)
E = Ease of use

Just for the sake of argument, let's say that F and V are 1-10, and E is 0-1, to give you a possible score of 0-100, a nice round number.

For example, take this popup window.

Some totally arbitrary scores for this popup window, off the top of my head:

F = 10 (it posts a comment, which is something I want to do)
V = 9 (I can clearly see that it will post a comment if there's no error)
E = .6 (It mostly works, but it can't be resized, which prevents me from seeing the whole comment)

This gives an overall score of 54.

The drawbacks are there, and annoying, but not enough to make me not use it.

Posted by: Adam Fields at July 31, 2004 09:30 AM

"Maybe what I'm saying is a bit extreme but we NEED something so the technophobes and illiterati stop ruining for the rest of us."

Hmmmm...yup...blame the users. Reminds me of hearing the comic Sinbad in Chicago at an early Web convention in '97 or so. He was speaking/riffing at the Apple booth. One great line comes to mind: When asked what he thought of the Web, he said, "Oh yeah...I know UNIX programmers who've killed themselves because NORMAL people can now use the Internet!"

Posted by: Joe at August 2, 2004 01:42 PM

As someone who works on a complex consumer-facing web application, "cutting scope" or reducing functionality as a path to simplification is not always an option.

Microsoft put out a paper a few years back that addresses the issue of complexity from a perspective that made sense to me.

Instead of immediately reducing the breadth of a product and therefore it's overall complexity, (or worse, assuming that the customer will, over time, "figure it out") Microsoft looked at ways to better manage the complexity of systems that are inherently complex. (Though it does not appear this approach has made it into any of their mainline products.)

The approach this paper describes is the Inductive User Interface. In short, this approach focuses on breaking down broad and complex systems into more manageable and, more importantly, digestible chunks.

They make a good case for what is often a difficult notion to communicate: That by increasing the "number of steps" (and arguably the complexity) you are making the product "easier to use".

More here...


Posted by: Mark Hines at August 2, 2004 02:30 PM

Danyel: WebTV wouldn't have worked quite like that in my dorm-- you have to dial 99 to get an outside line. I suppose whoever has extension 1800 would've been royally mad. ;-)

Posted by: codeman38 at August 2, 2004 06:37 PM

Oh so very very very very very true. Having just spent over a week upgrading my PVR to an high definition Tivo I spent hours and hours in total and complete frustration. To make matters worse the technical support for all of these devices is horrible. Yes I do have amazing and brilliant recordable high definition now but what a cost both in terms of real dollars and psychological ones.



Posted by: Thomas Hawk at August 2, 2004 10:19 PM

Hanan: Aren't 6) 'Does Not Require Parts' and 8) 'Is Modular' mutually exclusive?

Posted by: Jim at August 3, 2004 06:26 PM

I think you're basic idea is fatally flawed because the whole mess is just way too complex for any user to have even a fundamental grasp of how it works. Do we explain how viruses attack? Do they need to understand programming so they can grok buffer overflow attacks? Do users need to know something about dhcp and name lookups and NAT and firewalls? You want them to be responsible for what might happen to their machines, but you are basicly asking a sick person to become a virologist so they can avoid spreading the plague. Stuff should just work. The problem is that no one pays for usability. They buy features. So the market drives us all towards more complex, less usable stuff.

Posted by: jason rusoff at August 4, 2004 02:58 PM

More usable stuff is more expensive, so people gravitate to the cheaper, less usable stuff. Maybe the problem isn't that people won't pay for usability, but that usability has been defined as a feature that automatically carries a higher price, and people fail to see why they should pay more for a product that "does the same thing". Sure, maybe it's hard to do usability right (maybe even impossibly hard), but is that really a differentiating factor to the end user?

Why should they care what the development costs are? Typically, R&D costs (which should include design and usability) are factored into the early adopter rollout, then the price comes down. But usability has bee put on this huge pedestal, and it doesn't happen. Designers assume that they can collect on their "coolness" profit forever, and don't realize the extent to which they're alienating their core customers (who get written off as dolts who don't understand the value of good design).

Posted by: Adam Fields at August 4, 2004 07:08 PM

This is a great web site with some thoughtful opinions. Keep up the great work.

Posted by: Daniel Lenehan at August 12, 2004 09:00 AM

Jason Rusoff,

xeno's idea's are not as strange as you might think. Think about alle the licences and diploma's you need to do anything significant in this life? I can't drive a car unless I have my drivers license. If I want to fish in the stream that is practically paryt of my backyard, I still need to visit city hall and I need a piece of paper saying I can tie my shoes properly.

What's wrong with a computer diploma? Not one of those cheesy "I'm using teh intarweb!" things, but one that makes sure that the person using the computer knows what he's doing. For him/herself and others.

Computers nowadays have acsended the humble calculator status. I don't want my computer be totally messed up with endless wizards, help-balloons and happy-go-lucky animations everywhere. No amount of good usability is going to change that. Some people just can't get the hang of it. Just like some people have alot of problems with a stickshift, some people just can't type good enough. Or keep their emailadresses safe.

To continue with the car analogy, ask a random driver on how a turbocharged combustionengine works and 9 times out of ten he/she won't know how it exactly works, but that gasoline burns and an overheating enigne can go BOOM!, is all they know.

And need to know.

Posted by: Stefan Visser at August 15, 2004 12:25 PM

> Hanan: Aren't 6) 'Does Not Require Parts' and 8) 'Is Modular' mutually exclusive?

Nope. Good technology ships with all its necessary parts, so by the time you buy it, it requires nothing. But that doesn't preclude you upgrading or customizing it by buying or building different parts and swapping them in.

Yes, every product should have a big green button that says "go" -- but behind the scenes the button always kicks off a bunch of little processes strung together. You can't get around that reality, even though the button seems like one process to the user. Products I like are invariably those that allow me to go beyond the default green button behavior *when I want to*, by changing one little process at a time, and don't force me to do so too early.

... strangely, rules #6 and #8 are the only two (of the nine posted) that I really agree with. A few others, like "standardized" and "simple", are significant but not required. (How can it _be_ simple if it follows all those rules?)

Posted by: Travis Wilson at August 16, 2004 11:01 AM

Michal W. to stary gej

Posted by: r3d1p5 at October 26, 2004 04:20 AM

stasio mysli, ze smierdzicie

Posted by: stasio at October 26, 2004 05:10 AM

Własnie wruciłem z premiery "one Last Dance" i muszę stwierdzić że wilm jest kiepski a Swayze się zestarzał i ma strasznie przepity głos.....

Posted by: 5t453k at October 26, 2004 07:57 AM

I'd like to eliminate remote controls completely. I think a voice-activated system, right from the television, is the way to go.

I think harddrives, motherboards, chips, ram, and all things related to internal hardware, should be accesible through exterior bays and ports...No more screws, no more complex installations...Just plug and play.

I think that ViewSonic has the right idea. When I just got my new monitor, I opened the box, and pulled out the instructions...One sheet long, 3 steps, you're done...Of couse, we're talking a monitor, but the idea is what's important.

I think that networking should be pre-installed and pre-configured when you buy your router, modem, and I think the software and configurations should be pre-set-up and ready to go. There is a percentage of configurations that the majority of systems would fall under. The market should cater to them to eliminate confusion. While at the same time, be adaptable to the rest of us. In terms of corporate networking, this will always need consultants to build and manage.

Bottom Line: Simplify is key. It's never a question of technology. Instead, it always boils down to disorganizion and lack of future planning on the part of the manufacturer....Or is it? hmmm....Food for thought.

Posted by: Tom Levine at December 12, 2004 12:38 PM

Thanks for the special work and information! buy tramadol online

Posted by: caryl at April 4, 2006 05:03 AM