Every button, every word, every link, every switch is scrutinized to make sure it’s absolutely necessary and won’t generate a future support request.
With the obligatory disclaimer that I really love talking to customers even when they can't find the Save button: I agree with this so much it is physically painful to contemplate not doing it. There are thousands of them and one of me. If I screw up, there goes my weekend. (This has happened before at least twice.)
Relatedly, on the question of trusting what users do over what they say: if you instrument your site well (i.e. track events of significance), you can come up with your best two (or ten) candidates for a particular interface element, throw them both up in parallel to different users, and see which (empirically) is the easiest to understand. A/B tests: not just for signup forms.
It is insane how much mileage you can get out of this. e.g. I increased the percentage of customers who succeeded at one task from 50% to 60% with just a month of iterated A/B tests on two pages. It is a good way to cover for the fact that I have all the design/UX skills of a mole rat.
There's also a somewhat more difficult variant of this that I've been working on, which is supporting multiple interfaces at once and segregating users into them automatically based on classification of their skill level. It appears to work very, very well in my limited testing so far.
With the obligatory disclaimer that I really love talking to customers even when they can't find the Save button: I agree with this so much it is physically painful to contemplate not doing it. There are thousands of them and one of me. If I screw up, there goes my weekend. (This has happened before at least twice.)
Relatedly, on the question of trusting what users do over what they say: if you instrument your site well (i.e. track events of significance), you can come up with your best two (or ten) candidates for a particular interface element, throw them both up in parallel to different users, and see which (empirically) is the easiest to understand. A/B tests: not just for signup forms.
It is insane how much mileage you can get out of this. e.g. I increased the percentage of customers who succeeded at one task from 50% to 60% with just a month of iterated A/B tests on two pages. It is a good way to cover for the fact that I have all the design/UX skills of a mole rat.
There's also a somewhat more difficult variant of this that I've been working on, which is supporting multiple interfaces at once and segregating users into them automatically based on classification of their skill level. It appears to work very, very well in my limited testing so far.