The waterfall model is a software development methodology that follows a set sequence of actions from top to bottom, in the form of a waterfall, with its stages closely aligned to the software development lifecycle from which it evolved.

Where it came from

The first elements of the model were identified by Winston Royce in his 1970 paper on managing large software systems, though Royce himself described the idealised, purely sequential version as too risky for large projects and suggested iteration to make it more workable. The name "waterfall" appears to have first been used a few years later, by other authors describing Royce's concept.

How it works

Like a run of pools of water, where each pool fills until it overflows into the next, a waterfall project can only move to the next step once the previous action is complete and signed off. The stages are: define, where requirements and functional specification are agreed; design, where the technical specification is signed off; development, where the code is written; testing, where test cases are run and signed off; and deployment, where the application goes live and moves into support.

There's a high reliance on documentation throughout, particularly the functional specification produced in the define stage, which becomes the basis for everything that follows.

Where it still fits

Waterfall has fallen somewhat out of fashion because of its inflexibility, since defining the product's parameters at the outset doesn't lend itself to more modern, agile ways of working. But it remains a common methodology, and is often a first choice for large organisations and large projects where close control over stages and deadlines is genuinely desirable. Some teams use a hybrid, iterative waterfall, revisiting the immediately preceding stage where needed, though this can introduce its own log-jams if testing throws up a problem.

Benefits and drawbacks

Its strengths are real: a disciplined environment with detailed planning controlling every part of the project, clear timelines and milestones the whole team can see, and a detailed upfront specification that lets newcomers understand the project's direction quickly. Its drawbacks are equally real: changes beyond the define stage are costly and disruptive, testing happens late enough that major problems can jeopardise the whole project, and the entire model leans heavily on getting the functional specification right the first time.

Whether it's the right fit depends heavily on the nature of the project, and it's a conversation worth having early, before the sequence begins.