Evolving Your Web Site
Like this article? We recommend
Like this article? We recommend
Like this article? We recommend
After your site launches, the real work begins. For it's only after people visit your site that you can see what works and what doesn't. No matter how well you've planned, you can't always predict how a site will be used.
Successful web development, then, is an iterative process: You launch the site, study how people use it, and make continual changes to improve its performance.
This is the advantage of the web over other media: Everything can be measured, and then changed to reflect what's been learned. The measurability--and flexibility--of the web proves important to just about every industry, because it helps businesses understand their customers and better serve their needs.
"Quantitative research on the Internet is so luxurious," says Adam Berliant, a group manager for Microsoft, and previously a lead product planner for MSN.com. "I cannot pick up an issue of Newsweek and tell you which article was read the most. Or how much time people spent on each article. Or which headline was the most effective. I can't do that. On the Internet, I can! I can tell you what 100% of the people who used this site did."
But Berliant's interest isn't just academic. The purpose of studying your site is to figure out how to improve it. "I can run experiments and see which headline does better," he says. "I can take the same photo and run it with different headlines, or take different headlines and run them with the same photo--or whatever combination of things I want to test."
"You end up with this ever-building cumulative picture of what works and what doesn't work on a particular web site," Berliant continues. "And the beauty of the web over software is I can take that learning and act on it tomorrow."
studying your site
This is the one thing all successful web sites have in common: They pay close attention to what's happening on their sites. By watching what people do--where they come from, where they click, how long they linger--they learn what works on their sites, and how to provide more of it.
And unlike other industries, where data analysis is an arena for specialists, web companies spread this work throughout the product team so that everyone--designers, editors, engineers, producers--has the tools to make and analyze decisions.
"In a lot of companies, this kind of analysis ends up isolated in marketing," says Kris Carpenter, former VP of Excite. "The product team is handed some of the data, but they're not expected to be in the logs everyday. But when you have a product that's active and available online, suddenly the whole team needs to be familiar with the data."
And perhaps the most important thing a site owner can do is make this data available to their staff. This was always a priority for Beth Vanderslice, former president of Wired Digital (and the author's former boss). "I was always so passionate about traffic reports--and usability testing, and customer feedback," Vanderslice said. "Because if we weren't creating something that people wanted to use--or could use--then what's the point?"
"So the first thing I did was create a distributed reporting system, which allowed everyone in the company to log on and see how their sites were doing," Vanderslice explained. "I wanted to get that information into the hands of people who could use it--the designers, the engineers, the editors--so they could see for themselves what was working and what wasn't."
3 ways to study your site:
Traffic analysis. The first step toward understanding your site is studying your traffic. How many visitors do you have? Where do they come from? How long do they stay? This information--and much more--is stored in your site's log file, but must be analyzed by software to produce meaningful numbers. You can quickly move beyond the basics--getting a feel not only for how many people visit, but where they go, what they do, and why.
Usability testing. If traffic analysis tells you how your site is used, usability testing helps you understand why it's used that way. A typical test would involve watching a user as he performs a given task on your site. By watching what he does--does he know where to go? Can he correct mistakes?--you get a sense for how real users interact with your site.
Customer feedback. Although it may not always feel this way, customers pay you a big favor when they give you a piece of their mind. Email, letters, and phone calls offer valuable insight into who your users are, and what they want from your site.
Of course, people are more likely to write when they're angry than when they're delighted. So feedback tells you more about what frustrate users than what pleases them. When you do get positive feedback, it's a strong indication that you're doing something right. Keep in mind, however, that positive feedback doesn't always correlate with popularity. The things that inspire people to write a letter aren't necessarily the things that inspire use.
"The things we get the most email about are not the things that are most popular," says Esther Drill, co-founder and editor of gURL.com. "We get thousands of emails a day about some of our features. And they're definitely popular and they elicit contributions, but they're not the things that drive the traffic, not even close."
evaluating how people use your site
In order to learn how visitors--and how many of them--use your site, it helps to know what questions to ask, and where to find the answers.
How many people visit my site?
How do people find my site?
Where do people enter my site?
What parts of my site do people use?
How long do visitors stay on my site?
How often do visitors return to my site?
Where do people leave my site?
Where do people click? And why?
Who should I ask about this?
The log file (think of it as a guest book) can usually tell you the number of visitors. But accuracy varies, depending on the software you use.
The referrer file tracks the origin of each user who comes to the site. It reveals which sites sent you the most traffic, and how many users find you on their own (without following a link from another site).
Most traffic-analysis software can issue a Top Entry Pages report.
The log file keeps a running list of each page served from your site. Most traffic-analysis software will issue a Most Visited Pages report. And many will let you view pageviews by section.
If your site is application-based--and requires users to log in--you should have a log file that captures the length of each user session, in minutes. If your site (like most sites) doesn't require a log in, you'll have to estimate the length of visits by the number of pages each visitor viewed.
Again, if you have a site that requires users to log in, the log file should reveal great detail about repeat visits. Other sites can track repeat visitors by setting cookies but this method is less precise.
Most traffic-analysis software can issue a Top Exit Pages report.
A clickthrough report (available through some reporting systems) can track the number of people who click on any given link. Experimentation will tell you what factor (wording vs. color vs. placement) inspired the click. No one can exactly tell you why.
If you work for a large company that manages its own web servers, ask your system administrator what logs or reports are available. Otherwise, your hosting service should provide them. If your hosting service doesn't provide any traffic-measurement utilities, consider switching!
continual improvement is key
Continual improvement is the secret behind many sites' success. The big players, like Amazon and MSN, are always evolving. And smaller sites, too, benefit from this approach.
But learning to think of a web site as an ongoing experiment--rather than a finished product--is a big transition for many site owners. "In almost every other area of business--certainly in technology--the focus is on the product," says Andrew Anker, Internet pioneer and a partner at August Capital. "But on the web, it's really the process. The easy part is getting the product out. The real challenge is to figure out what you did right and wrong, and change quickly."
Put plainly, Anker says, "The biggest mistake you can make on the web is thinking you're done."