After last year's inaugural round of NSF Bio preproposals there was a flurry of discussion regarding what made a successful application. After attending a panel, I've expressed what I thought made a good preproposal and so have others.
At this point many of you should be well into the process of putting your document together. The DEB deadline is roughly two weeks away and the IOS one merely 10 days. Based on reading >30 of these documents in the last year, below are a few suggestions for jumping the bar to a full invite. These are things that I found made a proposal effective and piqued the most interest in the panel I was on.
1) Put your big question up front. I know, this should be a general grantsmithing point, but it is amazing how many people make the reviewer dig. The idea of these preproposals was to sell the big idea and whereas I address this below, the entire proposal needs to be wrapped around a good question that is clear from the start.
2) You need more than a big idea. The challenge of these documents is balancing the "sell" with the evidence of feasibility. My impression from talking to POs is that the initial intent was for these to be heavy on the sell of what a cool idea you have and the methodological details could come later. The reality I saw was what I would have predicted: scientists are skeptics by nature and don't want to be sold a bill of goods on empty promises. Don't look like a bridge salesperson.
3) Make it clear you can do what you are proposing. Again, this gets back to the issue of #2, but my panel saw a lot of people proposing to use cutting edge methods with which they had no demonstrable experience. That kind of thing just doesn't fly, no matter how good your idea is. Back up your methodology with papers, preliminary data or, if you have to, named collaborators. I strongly prefer showing preliminary data, even though I've had some POs suggest that wasn't expected. It worked for me and many others I saw get recommended for an invitation.
4) Use figures. It's tempting to cram your text in there because of the page limitations, but don't throw out the central tenants of grant writing just because you have less space. Summarize preliminary data, show how your aims interact, show a pretty picture of the cute animal you plan on ripping the spleen out of, whatever. Just don't offer up wall to wall text as a sacrifice to the gods of triage.
5) Provide a short methods section. You need enough to convince reviewers they want to see the Full Monty. Again, own the details in the context of the big picture - what are the key pieces of information the reviewers are going to wonder if you know?
If you can pull all this off in four pages of text, you are giving yourself the best shot to compete. The shortened format means that more people just tossed an idea out there and hope for the best. This favors the prepared, with at least 35% of the preproposals triaged early in the process and another 20% that were clearly non-competitive even after surviving triage.