Or: some of the few things I've learnt after a year of doing so and after giving feedback to people writing proposals too.
1) KEEP IT SHORT AND SIMPLE. This is the most important thing. Organisers often receive hundreds of submissions, and proposals will probably go through a first "screening" round. If your proposal doesn't sound awesome at a first glance, it's out. Done. End of the story.
2) Can this work as a talk? If it is a new subject, is it about Yet Another Framework that someone invented and has five users only? Would that make more sense as a blog post? If the subject is old and dull, can you add a different twist that makes it enticing and relevant? Maybe you can go a bit lyrical, but don't overdo it!
3) Why should anyone watch your talk? What's the expected outcome? Will it provide any specific value? Will people learn something? Will they be left speechless?
4) Don't add links. If you need to link to something else you haven't summarised enough. People reviewing proposals don't have time to click on links.
5) Spellcheck and copywriting. Does this have a coherent structure? Does it even make sense? Does it have typos? Sloppy writing always leaves a bad impression on the reader, even if they won't admit to it. Read it out aloud. Does it sound good? Can you read it with ease and not running out of breath? If you can't, go back to rewriting. Shorten sentences, work on the rhythm, polish it until you make it beautiful.
6) Give it to someone else to read, and listen to their opinion. Preferably give it to people from different backgrounds-they don't need to know the subject, but they can tell you if it sounds cool even if they have no clue about what you're talking about.