Obviously there are a lot of things to consider, but here are some general parameters.
Time Matters 10 is essentially a service release which fixes a large number of bugs and is generally more stable than previous versions. Also, the Pro version has been eliminated and replaced by the SQL version (previously labeled Enterprise). If you are a small firm running Pro, you should definitely upgrade to 10. It will be much faster and more stable.
After that, it depends on how much data you have (how long you have been using the program), whether you have heavily customized Time Matters and whether you are willing to “start over” (at least in part). Converting data to any other program will be quite expensive and probably incomplete (i.e., you won’t get all your data, especially if your TM has been extensively customized).
Next, of course if you are sufficiently frustrated to be willing to “start over” then you completely free to make any choice you want. Most people are not willing to do that.
Lastly, I agree with Steve Stockstill who posted a comment to the effect that it is not a question of “whether” the SaaS model will become dominant, but “when.” In my view, the current offerings are largely immature, but are developing very rapidly.
Therefore I think that the best option (depending on some of the above criteria) is probably to temporize: upgrade to TM 10, keep it for 2-3 years and then see where things are at.