This functions in many ways similar to crowdfunding, without the added overhead or stress of having obligations to fulfill. Explicit in the buystarter premise is that people are guaranteed nothing, they paid a discounted rate and are entitled to no improvements.
I have used this with two releases: "Under the Waterless Sea" and "The Price of Evil". "The Price of Evil" generated more funds than "Under the Waterless Sea", but neither had massive sales nor profits.
With that bit of context, my advice towards best practices is largely "Fudged if I know", since I only have two data points. But here are my educated guesses (which is a $10 phrase for "hunches").
1. Be frugal and decide if you really need art, if it adds anything to the works usability.
2. Layout is the most important thing to invest in.
3. You will not make nearly as much as you could with a kickstarter
4. You need to have a reputation of actually delivering, this might not go so well for a first release.
5. Don't dally, start working on anything you are going to do immediately, even when the buystarter is still going.
6. Avoid scope creep. You can always do more. Do that on a future re-release.
There you have it, six platitudes of limited use. Enjoy!