If software does everything it's supposed to do and it's bug-free, that's all you need, right? Well, not really. If users hate it and find it painful to use, you don't have a winner. Great software needs not just QA testing but usability testing.

It''s not enough for the product manager to test it. People who are familiar with an application won't see it the way a new user will. They'll miss problems. People who haven't seen it before need to try it. They should be the kind of people who'll actually use the software. Often more than one test group is necessary. If both administrators and data entry clerks will use the software, they''ll have different experiences, and you need testers from both groups.

Testing needs to start early in the process and continue through it. The developers need to get feedback, make adjustments, and then produce a new version for the testers.

Testers need to observe users and instruct them to think aloud as they're using the software. They should make notes on the users'' comments and non-verbal reactions; are they smiling or gritting their teeth? However, it's not just their reaction which is important but their performance when carrying out assigned tasks. Do they make consistent mistakes? That could indicate a badly designed interface. Does it take them a long time to finish the task? Maybe some reworking would make it easier.

Monitoring software supplements the observer's personal reactions with quantitative measurements. It can record the exact sequence of actions users took and show where they slowed down.

Usability testing doesn't have to be a huge or expensive deal. Research has shown that five subjects will catch the large majority of issues. But software that goes out the door without feedback from real users risks having to rework the finished project to make it viable. Make sure usability testing is part of your product''s development cycle.