Here's how I have explained it. And it really does depend on the industry you're in. Unless you are doing something that is very measuralbe, task percent complete is difficult to accurately report. Some people estimate based on the duration of time so that if I have a 10 hour task and I have worked 1.5 hours then I must be 15% complete. Some others go to the extreme and say that if a task is started it immediately goes to 50% complete, to mean In Progress. 0% obviously is Not started and 100% is complete. Best practice is to use the percent completes you listed above, although I've not seen the 95% used very often but I understand you to mean that it is almost done. This is why the % complete based on time doesn't always work well because it could take you 9 of the 10 days to get to 95% complete and then another 10 days to get it all the way to 100% complete.

here is an interesting article

http://rjmtech.blogspot.com/2006/01/how-percent-complete-is-that-task.html

Hope this helps

Amy,

Interesting insight.  When I am looking for task update metrics; I usually look for % complete on the work, and the duration only if I have scheduled things tightly. A 5 hour (work) task could be given 5 days of duration due to work conflicts. 

As long as the task doesn't exceed duration, I am looking for the actual time (work) spent on the task.  The resource may complete the task in 2 hours or 8 hours of work and can report % complete on the work.  If they complete the 5 hour task in 5 hours of work and 3 days duration, I still consider it 100% complete.  If they complete the 5 hour task in 20 hours and in 20 days, it is also 100% complete.  Duration shouldn't be a factor in % complete, it should be a factor in Schedule Variance because the actual finish dates will be different and will affect subsequent tasks.