Thursday, September 6, 2007

Form Submissions

Aaron Barker was just IMing me with some questions about form submission standards (such as they were) when I was at Sprint. We won't get into those, as they are muddied, and full of politics as well as design issues. He's got a similar issue, though, working in a big group, for a huge organization, trying to settle standards for it all. But one bit of info informing his question was a post by Luke Wroblewski on buttons. Its interesting, but I think flawed in some ways. Check out the heatmap. That must imply rigor. Well, its nice, but I'd rather have seen the numbers for speed to complete. Its mentioned in the text, but the article is overburdened with these graphics. Which, since they are what you'd expect, aren't that interesting. Overall, I am most disappointed with the dataset tested. I have always said that the primary design principles you can mess with are:
  1. Position
  2. Size
  3. Shape
  4. Contrast
  5. Color
  6. Form
(Form is like type weight, and other minor changes, but to familiar elements so its relevant. Still last though.) Many others have their own ideas. I am sorta enamoured of the time-based ones, and will have to ponder that a bit more. But working off my theory, items D, E & F changed position, but did not address the shape, contrast (minimally) or color aspects under test in A, B & C. Were it me, I'd have tested A-C, then re-tested with those results to make a set that tries out position changes independently. (Also, the use of that faded gray background under the buttons is way too specific an element for anything I'd test in this sort of research; I wouldn't go past space and an HR). But still, I have a problem with the A-C set. Gray is never a good secondary color. Gray is perceived (for real, in front of users) as unavailable. Grayed out. Users can work it out, but its not great. And, there was no consideration of shape (aside from the text vs. button) or size. Size doesn't have to mean there are two button standards; "submit" and "cancel" are basically identically long. I (and content people I have worked with) always tried to make the cancel action shorter, much shorter if possible, than the submit action. More like "Save changes" and "Back." This does help. Shape and color can be pursued really interestingly if you want to. Bank of America used to have these neat color/shape items accompanying their buttons. From memory, they were something like this. Sprint uses these little GT symbols for some buttons, and I occasionally tried to use the GT version on the submit, and nothing for cancel, but am not sure if it matters with the lack of much implication for the symbol, its frequent use as a bullet point across the site, and its relatively low contrast as a while element on a colored button. Additionally, forms are hardly ever as neat at those he's got in there. There's lots of discussion of forms going right down the left side, then you arrive at the buttons. But even this over-simplified version the country selector immediately above the buttons requires the user to jog over to the right. While there is mention of form F being least efficient due to fixation times, I see bounces left and right in all the tested cases. How much change? How much impact on total time? Could a different visual treatment of the page itself have changed how this worked? Overall, I've had very good results with the purportedly failed version E. Submit/continue on the right works well, especially for multi-step processes. But it does need to be consistently used, the graphic design of the page needs to support it, and secondary buttons need to be differentiable even without positional changes. Of course, I don't have access to Sprint's research, couldn't share it in detail if I did, and don't have a laser to shine in people's eyes to prove this, so what do I know?

No comments: