Limit the hours authorized to just long enough to figure out if they're worth their rate (2-3 is what I use), then up them only when you like a candidate
If a task is important, hire the best 3 candidates to do the first two hours' worth. Usually 1 will totally flake somehow, 1 will almost do ok, 1 will be good.
Weigh heavily on iterative solving (inspect their work daily (archived attachments are a must), interact, and don't give them too long of a to-do list)
Avoid skype and off-oDesk emails; it's always good to keep things in writing to maintain perspective
Recognize that they have other projects going on and that you can't occupy the lowest-bidder and highest-priority spaces at the same time (though one or the other isn't too much to expect)
*Don't be afraid to walk away and stop throwing good money after bad if performance suffers (usually that means they get unresponsive or slow to return work)
Hourly only, with work tracking.
Limit the hours authorized to just long enough to figure out if they're worth their rate (2-3 is what I use), then up them only when you like a candidate
If a task is important, hire the best 3 candidates to do the first two hours' worth. Usually 1 will totally flake somehow, 1 will almost do ok, 1 will be good.
Weigh heavily on iterative solving (inspect their work daily (archived attachments are a must), interact, and don't give them too long of a to-do list)
Avoid skype and off-oDesk emails; it's always good to keep things in writing to maintain perspective
Recognize that they have other projects going on and that you can't occupy the lowest-bidder and highest-priority spaces at the same time (though one or the other isn't too much to expect)
*Don't be afraid to walk away and stop throwing good money after bad if performance suffers (usually that means they get unresponsive or slow to return work)