user acceptance testing

hello,
can someone provide me with a general idea on what is to be tested?
a template would be great.
thanks.

Any kind of testing needs to be specific to the system itself, rather than being something that you can get from a generic template.


It's late here so I'm a bit tired, but what my mind is telling me at the moment is that the idea of UAT's is mainly to check 3 things:

• The system works according to the specifications (which the users should have agreed previously!)
• The separate parts of the system interact properly
• The specifications were actually correct (eg: there aren't things that the users need that were left out of the spec's).

To test the first, you should create plans based on the specifications themselves, but broken into small independent chunks (such as "add a record"), as the chances are that you will need to correct something in one or more sections, and re-testing (as well as fixing) should be easier with the smaller chunks.

You may find that the sections are interdependent at times tho, as cases like "delete a record" will require that there is a record that you can delete!


To test the second and third, you use a much larger plan - but before doing it you should generally have completed the first (as well as any corrections and re-testing that arise).

The idea of this part of the testing is to simulate the entire process that the system covers, which may be an hours worth of work, or it may be a months worth (if there are monthly reports, etc). In the majority of cases the workload can be much smaller than normal (eg: if data is entered 10000 times a month, you could test with just 200), but it does depend on the system.

do i have to create a test plan to evaluate the interface as well?

It should be tested - it has a big effect on usability, and the impression that the user(s) get of your work. Whether it needs a separate test plan is up for debate, and is dependent on your system (as it may be covered by other test plans, or be easily merged into them).


In an ideal world you would test every aspect of the system to destruction, but of course the users aren't just there to test, so you need to prioritise what they do, and live with the fact that they will probably only have enough spare time to do a small percentage of the work they normally would (thus less data in the system etc).

There may be issues that arise later which you could have found with the UAT, but to find them the users would have spent a lot of time looking at other areas that didn't have issues.

i've actually tested the interface in my system testing phase based on the usability heuristics. is that right?

That's a good move - but any user interface should be tested by the actual users too (assuming of course that there is a user interface!), as it is something that can be affected by opinion/previous experience/etc.

alright so i guess i'll let users test both the functionalites and interface for my uat.