The main numbers are usually how many procedures need to be digitized, how long and complex they are, how many sites are involved and how many people will use the system, split between those who write and approve procedures and those who run them in the field. The number of integrations and the level of support needed also matter.
Having rough estimates of these up front makes it much easier to get a realistic proposal, since procedure libraries vary a lot in size and complexity. If exact numbers are not available, a sample of procedures from one site is often enough to estimate the rest.