wige
07-03-2007, 02:22 PM
Usability testing is one of the most important aspects of designing a site that is effective and provides a positive return on investment. Users have to be able to navigate a site quickly and efficiently, find the information they are seeking easily, and complete the tasks or transactions they came for with minimal disruption, or they will leave the site for a competitor. Issues that interfere with the user's goals, however, are often not obvious to the people who have designed the site and use it every day. Technical users also tend to behave differently on a web site than the general public, and employees of a company usually interact with the company web site differently than they interact with the rest of the web.
Unfortunately, hiring a usability consultant to review the site can be very expensive, and setting up a formal usability test can be time consuming and also very expensive. The methods that I mention here are not a replacement to a formal usability analysis, but rather an inexpensive way to perform small tests with a few users to find the most critical usability problems. This process may not find everything a more formal test would, but missing a few things is much better than missing everything because no testing at all was done.
The first goal is to determine what you will be testing. For most sites, this is an easy process. If you have an informative site, you want to make sure the information you are providing is easy to find. If you are selling products, you want to make sure that your users can quickly and easily place their orders, and that they have can easily find the appropriate products. Beyond that, the goals would vary from site to site.
Next, you need to create a test environment. You need a private or semi-private area for the test, where the user can be seated in front of the test computer. Another seat should be set for the tester so that the tester can see the user's reactions to various areas of the site, as well as the screen. If other people will be observing, they should not be in the room with the user. Ideally, you would want to record each test with two cameras, one to record the user's face and another to record the screen, with any audience watching on CCTV. However, if you are on a budget, this will probably not be possible, in that case make sure you take copious notes.
For the test computer, I would suggest using MS VPC with the IE6 compatibility image, because this software is free. Also, VPC has a "save state" option that lets you quickly reset the operating system after each test. This will also let the user use your network connection to view the web site without exposing any secure areas of your network or letting the user easily access the underlying computer. Links to the various pages you want to test should be saved on the desktop or in the bookmarks for convenience.
Now you need to get a pool of users. There are several ways to do this, but the cheap way is to use friends, family and coworkers who do not interact with the site on a daily basis - this is especially good for the first few tests as you can practice administering the tests, and may find areas in need of improvement right off the bat. After this stage, test subjects can be recruited through an ad on the web site (from your existing customer base) or through advertising through local sources such as newspapers. This is where the real cost comes in as the users will need to be compensated, and you will have the cost of advertising to recruit the users. But the good news is you don't need to test dozens of users. A few users at a time can reveal a wealth of information that leads to a drastic improvement of your site. You should prescreen the applicants, making sure they fit into the target demographic of your site. Bear in mind the more selective you are in your screening, the more you may need to compensate the users. If you are recruiting general web users in your area, you may be able to offer discounts, or under a hundred dollars per users. If you have a web site that is geared toward investment professionals, for example, you may have to offer significantly more.
Finally, the time has come to test your users. In most cases, this follows the form of having the user perform customized tasks based on the process you are analyzing. For example, if you are testing the usability of the ordering process for a sports collectibles web site, find out from the user what sports they have an interest in, put them at an appropriate starting area (home page, landing page, etc, depending on what you are testing) and watch how they find the desired products and proceed through the order process. Make sure you encourage the user to think out loud and share their impressions, and let you know if they are having problems at any point. Also, keep the process interactive. If you see the user hesitate at a certain point, for example, ask what they are thinking. Try not to "lead" the user during the test unless they get completely lost and it is obvious they can't find their way back - your typical users don't have someone there to help them. But also make sure you keep track of where they got lost, and how they went astray. Also, try to limit the testing time to an hour.
After each user test, you will probably come away with a list of suggestions and ideas. After four or five tests, you will see patterns begin to emerge. These patterns will make the bulk of user issues apparent, and give you a good idea of changes that can help improve your site. But don't forget, each change you make will have it's own effect on the user experience. And that is another time to do a test. This is why small sample informal tests can provide such massive benefits - for a relatively low cost, you can gain a wealth of information that helps improve your site, and you can retest periodically to ensure your design keeps the user experience at the forefront.
If you have done usability testing like I mention here, or have additional suggestions or anything I missed, please let me know.
Unfortunately, hiring a usability consultant to review the site can be very expensive, and setting up a formal usability test can be time consuming and also very expensive. The methods that I mention here are not a replacement to a formal usability analysis, but rather an inexpensive way to perform small tests with a few users to find the most critical usability problems. This process may not find everything a more formal test would, but missing a few things is much better than missing everything because no testing at all was done.
The first goal is to determine what you will be testing. For most sites, this is an easy process. If you have an informative site, you want to make sure the information you are providing is easy to find. If you are selling products, you want to make sure that your users can quickly and easily place their orders, and that they have can easily find the appropriate products. Beyond that, the goals would vary from site to site.
Next, you need to create a test environment. You need a private or semi-private area for the test, where the user can be seated in front of the test computer. Another seat should be set for the tester so that the tester can see the user's reactions to various areas of the site, as well as the screen. If other people will be observing, they should not be in the room with the user. Ideally, you would want to record each test with two cameras, one to record the user's face and another to record the screen, with any audience watching on CCTV. However, if you are on a budget, this will probably not be possible, in that case make sure you take copious notes.
For the test computer, I would suggest using MS VPC with the IE6 compatibility image, because this software is free. Also, VPC has a "save state" option that lets you quickly reset the operating system after each test. This will also let the user use your network connection to view the web site without exposing any secure areas of your network or letting the user easily access the underlying computer. Links to the various pages you want to test should be saved on the desktop or in the bookmarks for convenience.
Now you need to get a pool of users. There are several ways to do this, but the cheap way is to use friends, family and coworkers who do not interact with the site on a daily basis - this is especially good for the first few tests as you can practice administering the tests, and may find areas in need of improvement right off the bat. After this stage, test subjects can be recruited through an ad on the web site (from your existing customer base) or through advertising through local sources such as newspapers. This is where the real cost comes in as the users will need to be compensated, and you will have the cost of advertising to recruit the users. But the good news is you don't need to test dozens of users. A few users at a time can reveal a wealth of information that leads to a drastic improvement of your site. You should prescreen the applicants, making sure they fit into the target demographic of your site. Bear in mind the more selective you are in your screening, the more you may need to compensate the users. If you are recruiting general web users in your area, you may be able to offer discounts, or under a hundred dollars per users. If you have a web site that is geared toward investment professionals, for example, you may have to offer significantly more.
Finally, the time has come to test your users. In most cases, this follows the form of having the user perform customized tasks based on the process you are analyzing. For example, if you are testing the usability of the ordering process for a sports collectibles web site, find out from the user what sports they have an interest in, put them at an appropriate starting area (home page, landing page, etc, depending on what you are testing) and watch how they find the desired products and proceed through the order process. Make sure you encourage the user to think out loud and share their impressions, and let you know if they are having problems at any point. Also, keep the process interactive. If you see the user hesitate at a certain point, for example, ask what they are thinking. Try not to "lead" the user during the test unless they get completely lost and it is obvious they can't find their way back - your typical users don't have someone there to help them. But also make sure you keep track of where they got lost, and how they went astray. Also, try to limit the testing time to an hour.
After each user test, you will probably come away with a list of suggestions and ideas. After four or five tests, you will see patterns begin to emerge. These patterns will make the bulk of user issues apparent, and give you a good idea of changes that can help improve your site. But don't forget, each change you make will have it's own effect on the user experience. And that is another time to do a test. This is why small sample informal tests can provide such massive benefits - for a relatively low cost, you can gain a wealth of information that helps improve your site, and you can retest periodically to ensure your design keeps the user experience at the forefront.
If you have done usability testing like I mention here, or have additional suggestions or anything I missed, please let me know.