Rapid Application Development is a software methodology in which software is prototyped and tested quickly through its development to arrive at a final product. It's one of several models that can sit within the broader software development lifecycle, alongside approaches such as Waterfall, Spiral, and Iterative and Incremental development.
Where it came from
Software methodology didn't really emerge until the 1960s, and a plan-driven model based on engineering practice seemed the logical way to go. As dissatisfaction grew with specifications defined upfront that were no longer valid by release, James Martin, who began his career at IBM in 1959, developed an approach called Rapid Iterative Production Prototyping, which led to his 1991 book on Rapid Application Development.
The James Martin model
The model runs across four stages. Requirements and planning, where business needs, scope and constraints are agreed jointly by users, designers and developers. User design, where users and analysts work together on prototypes for every process, input and output, iterating until users are satisfied. Construction, where the actual coding, integration and testing happens, with users still providing input as it proceeds. And cutover, covering installation, user acceptance testing and training.
It's a compressed version of the traditional lifecycle, offering a faster route to a fully developed, operational product that, thanks to continuous user input, more accurately reflects what people actually need.
When RAD makes sense
RAD works best when a project can be broken into modules, when users are genuinely available to help produce prototypes, when the budget allows for skilled engineers and design input, when domain expertise is close at hand, and when the customer is committed to change and can accommodate incremental delivery within small time frames.
Pros and cons
On the plus side: faster delivery, improved productivity, better cost and schedule control through timeboxing, reduced risk of the customer not getting what they want, continuous feedback, and requirements that can be incorporated as they change. On the downside: it only really works in a modularised environment, needs a highly skilled team and sufficient budget for modelling and prototyping tools, depends on constant user involvement, and can be difficult to fit within very large-scale projects.