How do you decide which cultural history sources to digitise first?

Our archive has far more material than we can scan in the next few years, and everyone has a different view on what matters most.

I'm weighing research demand against preservation risk and teaching use.

How do you set priorities when you can't digitise everything?