Comment by jdietrich | Hacker News Reader

Comment by jdietrich | Hacker News Reader

Parent Full thread

The important stuff in the common design pattern has been tested to death - everyone who matters is using clicktracking, a/b/x testing and deep user analytics. The reason they all look the same is the same reason that all infomercials or porn sites use the same basic structure - that's what testing indicates will convert best. Some people know this empirically, some are just blindly aping the fashion, but there is wads of data backing it.

Lots of whitespace improves readability, as does the use of a sans serif font. Deviate from either and a lot of people will hit the back button because they can't easily read your text.

A clear call to action massively improves conversion. The rectangular button in a dominant colour will increase signups by 10-20%.

960px wide because the majority of web users can see a page this wide without scrolling. For the same reason, all the important stuff on the site should fit within the first 500px vertically. a 960px grid is highly divisible and so gives you very flexible layout options with minimal hassle. There's a strong argument for 720px, but it's largely a question of browser demographics.

The links in the footer should be stuff for people who know what they're looking for, providing greater information density without added clutter. It works in large part because it's a convention, but it works nonetheless.

Familiarity matters in web design - the better people can predict where things will be, the better they can navigate.

That isn't to say that all these websites work as well as they should do, but there's only one reason to deviate from the norm - if you've tested something and found that it converts better.

You can only a/b test what you have come up with in the first place. This is not convergence on the optimum. It's just slightly optimized fashion.

Bingo! And who isn't today immune to infomercials? Or better yet, who doesn't unconsciously avoid big, shiny graphics with the belief that they're banner ads? (I know I do)

The point being that unified design will pick up a unified brand. The original poster is noticing the Young Startup brand, I suppose. In time this brand will become associated withall young startups and the enotional aftertaste will dominate a/b tested behaviors.

Thus the virtue of being innovative in design. Ita not going to (necessarily) lose on a/b testing and it'll help you to avoid brand unification.

As a minor footnote, sans serif DOES NOT improve readability. That's a myth. A nice serif typeface such as Baskerville is just as readable as Helvetica & friends.

On paper, serifs can be printed with enough resolution to aid in readability. On screens however, serifs are rendered too coarse and distract. That's why with current screen resolutions, sans serif works better for screen type.

I stand by my assertion that this is a myth. Maybe for small font sizes it matters, but on modern displays, on medium font sizes, in the age of sub-pixel rendering, it shouldn't.

I went Googling for science on this subject and promptly found a review of the usability literature:

http://www.alexpoole.info/academic/literaturereview.html

and a second study:

http://www.unc.edu/~jkullama/inls181/final/serif.html

And a poll of user preferences:

http://www.wilsonweb.com/wmt6/html-email-fonts.htm

The apparent conclusion:

Serifs hsve a small-to-nonexistent effect on the readability of digital text, one way or the other.

However, other qualities like x-height and letter spacing do make a difference. In a Times New Roman vs Arial battle, Arial tends to come out on top. But that may well be because it has a larger x-height.

When you poll users on their preferences they love sans-serif fonts. But a big part of the problem is that so few serif fonts come installed in all browsers. Times New Roman seems to be a poor screen font, and Georgia is better but isn't installed everywhere. In short, the fact that we prefer sans serif fonts for the web may be an artifact of our crappy selection of web fonts.

It will be very interesting to see if and when serifs stage a comeback in the era of 300dpi displays and, more importantly, a much larger choice of fonts.

This might all change with the entrance of tablets, especially if they reach higher resolution.

Serif fonts have the advantage that they have stronger row structure than sans-serif as the serifs act as a line.

So with the improvement of dpi in tablets happen, so will the use of serif fonts increase.

Wasn't this just on HN a few days back? Why you dont use Garamond on the web - http://www.kadavy.net/blog/posts/design-for-hackers-why-you-...

I find serif's easier to read if they're large enough (Readability does it right I think, fairly large serif text).

Indeed, and this trend also seems to prefer large-ish font sizes.

The Baskerville font you are talking about (Monotype’s) is based on large-ish sizes of a typeface designed in the 18th century for print. The stroke weight, x-height, and letter shapes are not well suited to setting blocks of digital text at normal screen sizes. It might work okay for titles and headings, but you should stick with Georgia or similar for body copy.

Helvetica is not an especially readable typeface for more than a few words. It works best for uses like logos, signs, posters, etc. For website copy, Verdana is a much better choice.

I'm interested in more detail regarding your comparison of Helvetica and Verdana, if you're feeling generous.

A lot of people will hit The Back Button if you don't use a sans serif font? Come on.

I'm willing to bet that the average person can't even tell the difference between Serif and Sans Serif typefaces. Not that this will have an effect on their self-conscious, but it's definitely worth something.

I think you mean subconscious.

Part of me wants to vote myself down. Kudos to that.

Don't make him self-conscious.

I actually wrote a blog post about this a while ago, called "imitation is the sincerest form of flattery" (http://goo.gl/K4RY) because I was hearing people complain that the site I was designing was too "similar" to other sites, specifically 37signals' product marketing sites. But who cares? They're good design patterns. You have to start somewhere.

And aside from that, building a promotional site is hard: http://goo.gl/8OlP. There's a lot of work there that's hard for a programmer to predict in advance, given that programmers generally haven't done things like think about how to market their product to an underserved niche. Tongue somewhat in cheek.

Reply on news.ycombinator.com