Assuming that all of your address space is private space there are several factors that you would want to keep in mind in order to allow for growth and scalability. Also, because you have the option of designing the subnets from the start you have the luxury of ensuring that there is a meaningful assignment of addresses. Some design methods include using the address space to help identify things such as regions, purpose, building, etc.
Example:
10.X[Purpose].X[Site]].X[Host]
10.100-150.10 - 19.X = Data Center at Site 1
10.100-150.20 - 19.X = Data Center at Site 2
10.10 - 19.1 - 128 .X = Data VLAN at Site 1
10.10 - 19.128 - 223. = Voice VLAN at Site 1
10.10 - 19.224 - 240.X = Printer VLAN at Site 1
This is obviously an example but you can use your address space to help identify meaningful information and allow for easy assignment, allow for scalability, and create predictable patterns which will greatly help any engineers that come on after the initial design or help those that may have to take up your place in your absence. The first thing that you need to do in designing your network is understand the infrastructure that you must support. Without knowing what infrastructure you have to support you cannot adequately design a solution to meet the needs of your network. From there it comes down to "best practice" which "should" be followed but may not always be applicable.
If this is helpful please make sure to let me know!
