The Old New Thing

This is why we get so many lousy user interfaces, because somebody decides that people are better off doing something in a prescribed way, rather than doing it in a more natural way.

To that end, I think e-books will never work, people prefer handling a paper version. Hypertext rarely works to my satisfaction (MS help documentation is often big offender in this area - and I keep sending them feedback) - I am often frustrated by jumping to a page with no more than a single sentence or paragraph.

You know, on shopping websites, actually buying something is optional.

I suppose you also object to being forced to put your name, address, contact details, etc... in to actually buy something. After all, if buying something is optional, then all of the fields in the checkout form should also be optional. You don't want none of that mandatory optional stuff there, neither.

OK, OK, I agree it's stupid to have *those* 43 questions be mandatory in *that particular* feedback section. e.g. you may not have used half the site they're asking for feedback on, or something. So that could probably be improved.

But to complain about *any* mandatory fields in optional parts of a website? I think you're taking that one a little too far.[0]

Interestingly, I wonder if you're required to fill in something in the freeform text boxes also. After all: "You must complete all the feedback sections before submitting your feedback." What if you don't have anything to add? Must you put "N/A" or similar to complete the form.

[0] Conversely, I hope this isn't going too far and earning me a place in nitpickers' corner. I think it's a legitimate counterpoint to your main idea. *crosses fingers*

Adam:

Last time I brought a counterpoint with a analogy[0], Raymond yelled at me! Like a girl, ok, but still!! :)

Good luck with yours!

[0]: And the analogy wasn't mine but his! Your case is so much worse I will have to light up a candle in a church for you... much more powerful than fingers crossed! :)

On shopping sites signing up for an account should be optional. If i buy something in a store, i dont have to pick a username and password that im never going to use again, and will probably forget anyway. Billing information is obviously mandatory.

For example, this weekend i was purchasing movie tickets online. Why do I need to sign up for an account? The tickets are non refundable, there isn't any shipping to track, etc. If the theater wasn't far nicer than any other nearby i would have made purchasing the tickets optional as well.

A few sites allow you to purchase items without having to sign up for an account which i really like for one time purchases. If its a store i frequent, then an account is more acceptable, but i dont want it required that i have a login if i dont intend to come back.

So yes, it actually does mean buying something is optional to me.

Well, in case of a user-interface the preferred way is to have just mandatory fields. Of course, every fields will be needed in the further process.

This happens more than half the time:

When installing XYZ, I am presented with a pre-checked option to "Show me the advantages of ABC".

If I forget to turn this off and click the installer's "Finish" button, the dialog goes to lunch and never returns.  I have to kill it manually via Task Manager.

So here is a case where the "optional" option is defaulted ON for me (on a MANDATORY download) and results in a computer near-death-experience unless deactivated.  

Irony aside, what possible reason is there for the checkbox?  Is this another example of the marketing suits hovering over the shoulders of the dev team, or a legal issue, or what?

The option was to provide all the feedback they wanted or none at all.  It's just common sense man!

Ha, you're lucky, more often that I'd like to see, when I click a survey button all the work that went into the form for which the survey was asking about get's lost.   I no longer fill out web-based survey's like that.

I've been harking on about this for a while. This happens all the time, even on Microsoft surveys. I would be willing to bet they would get a lot more feedback if they removed mandatory fields.

This reminds me of a recent experience with my bank. I received email asking me to fill out an online survey. I usually ignore these things, but this time I decided to do it -- I really want to give them specific feedback on their online bill paying services.

Several questions into the survey, I hit the following gem (paraphrased):

Have you tried online statements instead of paper statements?

(O) Yes, I tried it, but I went back to paper statements.

(O) No, I haven't tried it.

Note: There is no answer for the "yes, I tried it, I'm satisfied with it, and I have no need for paper statements now", which happens to be my situation.

Choosing the "yes..." answer leads to a series of questions probing for reasons online statements didn't work out. Choosing the "no..." answer leads to questions for reasons online statements haven't been tried yet.

Of course, one MUST choose one of these to continue the survey.

Lame.

My solution would have been slightly different from Raymond's.  I would have quickly marked all of the 43 questions with "completely unsatisfied" or whatever the lowest score could be and then give them my verbose feedback.  I like to operate under the philosophy of "ask a stupid question and you will get a stupid answer".

I think what they meant with optional is that the survey is optional. However inside the survey none of the questions are optional.

yeah they could have worded it differently

Heh, reminds me of a sequence of prompts I saw in an installer for a major programming suite. It went roughly like this:

Prompt 1: Your computer needs to be rebooted before you can use the installed program. [Reboot now] [Reboot later]

I picked "Reboot later", because I planned to install a few other packages before the reboot.

Prompt 2: Setup has failed to complete as you chose to not reboot the computer. You must reboot the computer before you can use the product. [Exit setup]

Way to go, scaring the user. Ah well, clicked "Exit setup", and was presented with this gem:

Prompt 3: Setup will now reboot to complete the insall. [OK]

Wait, didn't I just tell it to not reboot?. I wonder if the "Reboot later" path was ever tested.

I totally agree - I had almost the same experience a while ago.  There's really no reason all N questions of a survey should be mandatory...if I had really wanted feedback to give I would have been tempted to just make up answers to all the other questions.

Of course, part of the problem is that they felt the need to ask you 43 separate questions (I assume you're not exaggerating).  It's common for people to say "What is this, twenty questions?" when people get too inquisitive; of course that's a reference to the lame childhood game, but beyond that, it shows quite clearly that 20 is excessive to most people.

I wouldn't expect good, well-considered answers to more than, say, 5 questions (maybe 10, if the product/service offering were immense in scope).  Any more than that and people treat it like a quiz, just trying to get through it as quickly as possible.

The funny thing is, you just know that they're using the survey to settle some internal corporate dispute.  There's no other reason to get so specific.

Maybe the explanation is simple: they don't really want to get feedback, because they were expecting mostly negative feedback! Probably their boss insisted on a feedback form, so they implemented one but made it as hard as possible to send feedback.

Negative feedback is often more important than positive feedback.  People are more likely to complain to others about your products/services than they are to shout your virtues.  On the other hand, they're more likely to tell you they liked it (more likely than telling anyone else).  This doesn't help you as much as if they complain to you instead of their friends.

"The option was to provide all the feedback they wanted or none at all.  It's just common sense man!"

Absolutely not. My prediction is that they'll get so little information back that it'll be statistically insignificant. If they made all fields optional at least users who only cared for a small part of their app (positively or negatively) would be able to tell them what worked and what didn't and they'd get something to work with. E.g., which part of the survey was actually useful... :-)

As an aside: a survey with 43 questions for determining user satisfaction is obviously designed by someone so far removed from the user experience as to be totally useless anyway. I never bother with anything above 3 questions: "Are you generally satisfied with your experience", "What worked", "What could be improved". In my company we don't even have more than 10 questions on internal surveys after rollouts, it simply doesn't make sense to ask for more details from a large user group when doing this kind of measurement.

Today I ordered a bunch of bike jerseys (complete with the Rolling Stones tongue logo) and after filling out all the forms it finally objected that there was an error. Turns out that one of the items I had ordered was almost out of stock. Only one left!

After correcting my faux pas I once again had to fill in my address and credit card details as penance.

Yes, this is really annoying. I like to give feedback, but when I run into one of these I just opt out.

On a related note, <red>XXX</red> makes a couple foolish mistakes with their annual survey. They have low-end, midrange, and high-end products. If you are on a support contract for one (1) of them they send the appropriate survey for that product. My company uses the high-end product, but also has the legacy low-end product on support. They send only the low-end product survey, which is twice wrong.

1) They should send both surveys. If I don't feel like completing both, I might opt to answer the one about which I have something to say.

2) If they are only going to send one, wouldn't one assume the high-end product is more important to me? I certainly pay more for it! Or does this signal that the low-end product is more important to them, and I am about to be abandoned by a company opting to move down-market?

There sure are a lot of people who comment on this blog with no sense of irony whatsoever.  

There's a whole lot of websites that don't think nearly hard enough about their interface.  And presumably don't test with real people.

It is really dumb to lock up an input form with so much "error" checking that half the people won't ever complete it.  You'll never even know why.

Feedback.eedback.edback.dback.back.ack.ck.k... ;)

Leaving aside the bad UI, what puzzles me is that they think they'll get any kind of meaningful feedback out of those numerical ratings.

Say "Selection" gets a 2.  So they need more products.  But of what kind?  Their tedious survey will discourage most people from telling them what kind.

Of course, it's a lot of work to manually go through people's specific responses and translate them into action items, even if it has the distinct advantage of actually being effective.

It's pretty easy, on the other hand, to have a computer tabulate survey results, and then send out a mass company e-mail: "Attention all employees: We're running a 1.5 average rating on Friendliness.  All employees are hereby reminded to be friendly.  Employees refusing to comply will be summarily fired."

Actually, what irks me most about some online shopping sites are those that ask for your billing information (not just address, but credit card info) first - before telling you the total with taxes, shipping, etc.

And they usually give no indication that you will see the total before you submit - as if they may show you what they really charged after charging you.

Great post, Raymond! It seems like the moral of the story is that the only things that should be required to perform a task in a UI are things inherantly necessary to the task (ok, that principle can get violated with metaphoric interfaces, but at least it's for the good cause of learnability). Purchasing something on the internet pretty much implies providing a form of payment, an address, etc. Providing feedback does not imply answering 43 questions. It would be like going to purchase a copy of Raymond's book requiring you to also purchase 42 related titles!

"To that end, I think e-books will never work, people prefer handling a paper version."

No I don't.

I prefer my Sony Clie TH-55 over dead tree anytime. Dead trees don't have backlight.

to read Raymond's blog is optional...

to reply is optional...

to be flamed by Raymond for a reply is mandatory...

I've seen some tool that after installing offered me the choice between 'reboot now' or 'shut down without rebooting'. (!)

The latter allows you to continue. (!!)

I don't remember which tool it was, because I've had to reinstall a lot of them after an unlucky system restore.

......

As for the mandatory questions, they should make the whole survey optional (as RC said), but making each individual question optional (as Morten said) would probably not be a good idea. Just a single critical field skipped could render the whole thing useless.

Advanced statistic techniques exist for salvaging surveys with missing questions, but those surveys are already biased by a nonrandom sample, too much missing data would remove any remaining credibility. Skip the wrong two questions and your survey is garbage bin fodder.

The worst experience I had with a mandatory field was in an insurance company online quote web app.  After carefully filling in pages of information it refused to accept the quote because I 'hadn't entered a valid email address'.  I went back over the 5 pages of data entry; nowhere was there a field for entering my email address!  I phoned the company and told them; they hurriedly glossed over the omission but took my details for the quote over the phone there and then, at least redeeming themselves a little.

Yes, this is just common sense.  It was optional for you to click the feedback button.  After you have done that you can't pick and choose what feedback you want to provide.

PingBack from http://polymathprogrammer.com/2007/07/20/making-tough-programming-choices/