Presenting work in progress on conferences is a good way to elicit a first discussion of your (preliminary) research results, to establish yourself in a new topic, or to claim first ground in a "hot" research area. The important thing to me seems to be that you are able to present at least some results. Research concepts and plans often do not yield good conference papers.
There's a number of ways in which results can be regarded as premature, such that you would want to present them as "work in progress". I'll give some of them without claim of completeness:
In empirical research:
- Small amount of data compared to the standard in your field.
- Contradictory results which prevent firm conclusions.
- Low amount of replicates or additional supportive data for your conclusions.
- Experimental setups which may not capture or control for relevant factors in the system you're looking at.
In theoretical research:
- Restrictive assumptions which make a theory applicable to only a fraction of the relevant problems.
- Evaluation of a method with a smaller or more restricted test case than what would be done usually.
- Mathematical results from numerical experiments without a clear proof.
- Missing steps in an implementation such that only partial results could be obtained.
In many cases, it is a difficult decision, whether you should publish preliminary results (in case they are publishable in principle) or rather wait until you get more complete results. It depends on how fast you want to publish something on the topic, whether you can quickly get more complete results or not, and probably a range of other factors which I didn't think of now.