Skip to content
DT Nova

Software development methodologies: how to choose the right approach

Software development methodologies are the ways teams organise the work of building software: how requirements are gathered, how work is split up, when the customer sees results and how quality is checked. The main software development methodologies are plan-driven approaches such as waterfall and the V-model, iterative approaches such as the spiral model, and agile approaches such as Scrum, Kanban and Extreme Programming, often combined with DevOps practices. This guide explains each one in plain terms, where it fits, and how to choose a software development approach for your own project.

It is written for the people who buy or sponsor software, not only for the people who build it. The approach a team uses decides when you see working software, how changes are handled and what a fixed price really fixes.

You will also see these called software development models, especially for the plan-driven ones: the waterfall model, the V-model and the spiral model. Model and methodology are used for the same thing in most writing; the comparison below covers both.

The short version

  • Plan-driven approaches decide everything up front and build in stages. They suit stable, well-understood requirements and work where every stage must be signed off.
  • Iterative approaches build in repeated cycles, tackling the riskiest parts first. They suit large projects with real technical uncertainty.
  • Agile approaches deliver working software in short cycles and change the plan as people learn. They suit most business software, where users only know what they need once they see it.
  • DevOps is not a planning method but a set of practices that make frequent, safe releases possible, whatever the planning method.

Waterfall: one stage after another

The waterfall approach runs the work as a sequence: requirements, design, implementation, testing, deployment and maintenance, each stage finished and signed off before the next begins. It is usually traced to Winston W. Royce's paper "Managing the Development of Large Software Systems", presented at IEEE WESCON in August 1970. Royce did not use the word waterfall, and the paper itself argued for feedback between the stages and for building an early version to learn from, which is often forgotten.

Where it fits: fixed requirements, such as a system built to a published standard, software tied to a hardware schedule, or contracts where the scope must be complete before work starts. The risk: users see working software only near the end, so a misunderstanding in the requirements is found when it is most expensive to fix.

The V-model: testing planned alongside each stage

The V-model is a plan-driven approach in which each design stage on the way down is paired with a test stage on the way up. Requirements are paired with acceptance testing, system design with system testing, detailed design with integration testing, and code with unit testing. Test plans are written at the same time as the design they verify.

Where it fits: regulated and safety-critical work, such as medical devices, automotive and railway systems, where traceability from each requirement to its test is required. The risk: the same as waterfall, with more documentation to change when requirements move.

Iterative and incremental development

Iterative development builds the system in repeated cycles, each one refining the last. Incremental development delivers it in usable slices, each adding features. Most modern approaches combine the two: build a thin working version, put it in front of people, then extend and improve it. The spiral model, described by Barry Boehm, adds explicit risk analysis to each cycle: the team identifies what is most likely to go wrong and tackles that first, with a prototype if needed.

Where it fits: large or technically uncertain projects, where it is better to find out early whether an integration or an algorithm will work. The risk: without a clear end goal, cycles can continue without converging.

Agile: working software in short cycles

Agile is a family of approaches that share the values of the Manifesto for Agile Software Development, published in 2001. Its four statements are: individuals and interactions over processes and tools; working software over comprehensive documentation; customer collaboration over contract negotiation; and responding to change over following a plan. The manifesto adds that there is value in the items on the right, but more value in the items on the left. It is a set of values, not a process, which is why there are several agile software development methods: Scrum, Kanban, Extreme Programming and lean software development are described below, and many teams combine them. For how these ideas apply to phone apps, with app store review and releases in the loop, see our guide to agile mobile app development. For a direct comparison of the two, see agile vs waterfall.

Scrum

Scrum is the most widely known agile framework. According to the Scrum Guide, in its November 2020 version, work happens in Sprints, fixed-length events of one month or less. A Scrum Team has three accountabilities: the Product Owner, who orders the work by value; the Scrum Master, who helps the team use Scrum well; and the Developers, who build the product. There are five events: the Sprint itself, Sprint Planning, the Daily Scrum, a 15-minute event for the Developers, the Sprint Review and the Sprint Retrospective. There are three artefacts, each with a commitment: the Product Backlog with the Product Goal, the Sprint Backlog with the Sprint Goal, and the Increment with the Definition of Done.

Where it fits: a product team working on one product over months, with a business owner who can make decisions every Sprint. The risk: ceremonies without substance. If nobody with authority attends the Sprint Review, Scrum becomes meetings.

Kanban

Kanban organises work as a flow rather than in fixed cycles. Work items move across a board with columns for each stage, and each column has a limit on how many items can be in progress at once. The limits expose bottlenecks: when testing is full, nobody starts new work until something is finished. Teams measure lead time, the time from request to delivery, and improve the flow step by step.

Where it fits: maintenance and support, operations work and teams with a steady stream of requests of different sizes. The risk: without a product goal, a Kanban board can become an endless queue of small tasks.

Extreme Programming

Extreme Programming, XP, focuses on engineering practices that keep code easy to change: test-driven development, pair programming, continuous integration, small releases, simple design and frequent refactoring. Many teams that plan with Scrum or Kanban build with XP practices.

Lean software development

Lean software development applies ideas from lean manufacturing: eliminate work that does not add value, deliver fast, build quality in, and defer decisions until the last responsible moment, when the most information is available. It is less a process than a lens for deciding which steps a team can drop.

DevOps and continuous delivery

DevOps joins building and running software into one responsibility. In practice it means automated tests that run on every change, automated builds and deployments, infrastructure described in code, monitoring in production and the ability to roll back quickly. Continuous delivery means every change that passes the automated checks could be released at any time. These practices work with any planning method, and they are what makes short iterations safe: if releasing is risky and manual, nobody wants to release often.

Software development methodologies compared

ApproachRequirementsCustomer sees working softwareBest fit
WaterfallFixed up frontNear the endStable, fully specified scope
V-modelFixed up front, traced to testsNear the endRegulated and safety-critical work
Spiral and iterativeRefined each cycleAfter each cycleLarge projects with technical risk
ScrumOrdered backlog, changes each SprintEvery SprintProduct teams with an engaged owner
KanbanContinuous queueAs each item is doneMaintenance, support and operations
XP and DevOpsAnyFrequently, in small releasesEngineering practices for any of the above

How to choose a software development approach

Choosing a software development process methodology, the way the work will be organised from first idea to release, depends less on fashion than on four questions about your project:

  1. How stable are the requirements? If they are fixed by a standard, a regulator or a contract, plan-driven work is reasonable. If users will only know what they need when they see it, choose short iterations.
  2. How available is the decision maker? Agile depends on someone who can decide priorities every week or two. Without that person, iterations stall.
  3. How is the work paid for? Fixed price needs a fixed scope. That does not rule out agile, but it means fixing the scope per phase rather than for the whole project.
  4. What happens after launch? Software that keeps changing needs automated testing and releases from the start, or every later change gets slower and riskier.

Most business software projects end up with a hybrid: a clear goal and a written scope for the first phase, delivered in one- or two-week iterations with a demo of working software at the end of each, automated tests and releases underneath, and the next phase planned once the first is in daily use. The methodology is only part of the decision; who does the work matters as much, and our guide on how to choose a software development partner covers that side, and software development steps explains the stages every method passes through.

Software development life cycle methodology: stages and methods

In software engineering, the SDLC (software development life cycle) names the stages every piece of software passes through: planning, requirements, design, building, testing, release and maintenance, and finally retirement. A software development life cycle methodology decides how those stages are arranged in time. Waterfall runs each stage once, in order. The V-model pairs each design stage with the test stage that checks it. Iterative and agile methods run a small version of every stage in each cycle, so planning, design, building and testing happen again every one or two weeks. DevOps joins the last stages to the first, so release and operation feed straight back into planning. The stages do not disappear in any of them; what changes is how often each one happens and how much is decided before building starts. Each stage is described in our guide to software development steps.

How we work at DT Nova

We use that hybrid on client projects and on our own products. Each phase has a written scope and a fixed price agreed before work starts, so the cost is known. Inside the phase, we work in short iterations, and you see working software on your own data at the end of each one, not a slide deck. Changes you ask for are swapped against something of similar size in the scope, moved to the next phase or billed by the hour, agreed in writing. Tests and deployments are automated from the first week, and the source code sits in a repository in your company's own account from day one, so you can see every change as it happens. For how this plays out on a build, see custom software development; for choosing between building and buying, see software consulting; and for what is changing around projects in 2026, see software development trends and challenges. For the technologies a project is built with, see our guide to software development technologies.

Questions people ask

What is the most common software development methodology?

Agile approaches, Scrum and Kanban above all, are the most widely used in commercial software today, often mixed with DevOps practices for testing and releases. Plan-driven approaches remain common in regulated and safety-critical work, where documentation and sign-off at each stage are required.

Is agile always better than waterfall?

No. Agile works best when requirements will change as people see the software. Waterfall works when requirements are stable, well understood and must be signed off before building, for example when a regulator or a hardware schedule fixes them. Most real projects combine the two: a fixed goal and budget, delivered in short iterations.

What is the difference between a methodology and a framework?

The words are used loosely. A methodology is usually a broad approach to organising the work, such as agile or plan-driven. A framework, such as Scrum, is a specific set of roles, events and artefacts that puts an approach into practice. Practices, such as test-driven development or continuous integration, sit below both.

Can a fixed-price project be agile?

Yes, if the scope is fixed in phases. A common pattern is a fixed price for a clearly described first phase, delivered in short iterations with a demo at the end of each, and the next phase priced once the first is in use. That keeps the cost known while still letting the plan change as people learn.

Which methodology should a small company choose?

Usually a light one. Short iterations of one or two weeks, a single prioritised list of work, a demo of working software at the end of each iteration and automated testing cover most of what small teams need. Heavier frameworks add value as the number of teams grows.

Sources