It may help you do a rough first sorting. You can always adjust later in the process. Or you can rate all projects the same initially and do a completely customized allocation directly in the ballot.
The rating is only supposed to be a tool to help you; if you don’t think it’s helpful, you don’t have to use it. (I agree that this is not entirely intuitive, but it is my understanding from last round).
You can read more about it here:
Don’t cheat. 
I agree. I think diving at least somewhat into all categories is necessary. You need to have at least some idea of the overall size and quality of all contributions to do a proper job at budgeting.
You may discuss with other voters and use their experience too - are you a guest voter? You guys have a telegram group of your own, right? Maybe try to come up with ideas for allocation methods, and then discuss with the others. (We are encouraged to not share voting strategies between the two groups of voters - citizens and guest voters - which I find sad, but that is the way the experiment is designed, so…)
I would start by posting my question in my voter telegram group. Someone else may know something. Or have ideas as to how to handle applications with unclear impact statements.
If you think there is a technical issue with the application, maybe tag Jonas. But if it’s just that the applicant is unclear, try instead to come up with what you think is a fair approach to that. (Some might decide that if there is no clear impact, then they allocate 0; others might do extra research and find that there is an impact that should be rewarded - different voters, different voting strategies).
Whatever you end up doing, I hope you share your voting strategy afterwards, here in the forum. There will probably be a dedicated thread for it. You ask good questions - I’m sure it will be interesting to read about your answers and/or remaining open questions by the end of the experiment!