Normally, you would take a top-down approach in designing a LAN/WAN. This means that you first analyze the applications that you will use across the WAN, and determine the types of sites you will be connecting together. For examples, if you have a mission critial application from branch sites that must connect to central sites, and requires X amount of bandwidth, then you know you must get a link of X type. When cost is a major factor, you have to weigh the cost/benefit of the available connectivity options -- high price/high reliability vs. lower price/lower reliability. What is the cost to the business if the application can't perform it's function? There's a lot to consider.

Basically, the applications and availability requirements will dictate what kind of WAN you plan initially, and the type of WAN connectivity available in your area will dictate what kind of WAN you finally design/implement.

HTH