As part of my site redesign effort, I had to determine what the best screen resolution to target was. Initially, I was ready to blindly accept 1024x768 as "the standard," firmly believing that 800x600 has been left in the dust more than 5 years ago. Then I reminded myself that analytics have the real answer, but not just any analytics.
Despite the wealth of interesting information we can find online about the "average" screen resolution, the best answer is a little closer to home than these general studies. Chances are, your user community isn't the entire world. If you think it is, you don't know enough about your users. This is where analytics come in. For instance, the average screen resolution may be 1024x768, but if I'm building a site/application for the visually impaired, the average may be much lower -- maybe even 640x480! On the other hand, if you're building a site/application for gamers, the average will most likely be higher. In either case, missing the key demographic will undoubtedly provide a less than desirable experience for your users. This is where your own instrumentation/analytics come in.
I found out a few interesting things about my site's visitors from this exercise.
1024x768 is the smallest resolution. I should've expected this, knowing my audience; but I didn't. I figured there'd be at least a few 800x600 visitors out there. While I'm sure there are some in the world, this at least confirms that 1024x768 is the smallest resolution I need to concern myself with; however, at 11%, it's still a very small part of the overall audience.
26% use 1280x1024. While not that large of a percentage, the largest group of visitors has a 1280x1024 resolution. I was surprised to see this as the most popular resolution, but it makes sense, given hardware advances. Arguably, this is where my focus should be. A 1280-pixel width would support 87% of my visitors. Unfortunately, a 1024-pixel height only supports 59%, so that's not quite ideal. Luckily, this is just about scrolling. The best height to target for the prime real estate would be an 800-pixel height, which would support 86% of my visitors.
58% have widescreen displays. Again, not stunning, but very interesting. Knowing more than half of my visitors have widescreen displays tells me I should embrace the horizontal spread of key information, as opposed to dropping this info below "the fold." I'm not saying every layout decision needs to be based on "the fold," but it does exist and it should be acknowledged to a certain degree. Yes, people scroll, but the initial impact is above the fold, so that should be targeted and drive interest in not only scrolling down, but clicking thru to other pages. But I digress... The biggest lesson here is that fixed-width layouts are not an option. Limiting to a 1000-pixel width, for instance, would waste 25-50% of the display size for half of my visitors. That's crazy.
Most well known aspect ratios are the rarest. While 4:3 and 16:9 are the aspect ratios I hear about the most -- likely because they are the standard aspect ratios used in TVs -- they only make up 22% of my visitors. While knowing this won't help me come to a solid conclusion, it is another interesting point I wouldn't have expected. I only point this out because it shows how instrumentation can tell you things you may not have known you didn't know.
Strange resolutions like 1152x864 and 1366x768. I knew about most of the resolutions I found, but there were a couple I've never heard of, specifically 1366x768, which was very odd and doesn't exactly map to 16:9, but seems to be attributed as such.
There are a countless ways I could've sliced the numbers to extract other interesting tidbits, but the important thing is that I was able to use actual metrics to validate my assumptions. This would be even more important on larger projects; especially, team projects. Have you ever had an argument over some subjective aspect of a system you've built? Maybe it was a particular class design or visual layout or perhaps it was something as trivial as a color. We all have our opinions and design reviews tend to bring them all out. It's about time we stop having subjective arguments and actually pull together objective, factual metrics based on real system usage. What do your customers deserve? Assumptions and guesses or educated conclusions based on proven facts?
Source code for resolution visualization