Skip to content
Tutorials Center
My learning

Topics

Lesson 8 of 50 ยท The SDLC and Process Models

Choosing a Process Model

Learn how project size, requirement volatility, regulation, risk and team shape the choice between waterfall, V-model, incremental, spiral, Scrum and Kanban.

You now know several process models: waterfall, the V-model, iterative and incremental development, and the spiral model. Agile methods such as Scrum and Kanban come later in this tutorial. A natural question follows: which one should a given project use? There is no universal winner. The right choice depends on a handful of project characteristics, and getting it wrong is expensive: a rigid plan applied to a fast-changing product wastes months, and a loose process applied to a safety-critical device produces something nobody can certify.

This lesson lists the factors that matter, turns them into a decision table, and ends with a small JavaScript stand-in that recommends a model from your answers. The stand-in is a teaching aid built from simple rules of thumb, not a scientific instrument; real choices need judgment and often a hybrid.

The factors that matter

Think of these as dials. Each one pushes you toward some models and away from others.

Requirement stability. Are the requirements known and unlikely to change (replacing an old payroll calculation with an identical one), or are they a guess that users will correct as soon as they see something (a new consumer app)? Stable requirements allow plan-driven models. Volatile requirements demand short feedback loops.

Project size and cost of failure. A weekend tool and a hospital system are different animals. Large projects need more coordination and documentation; projects where failure is very expensive need risk analysis up front.

Regulation and evidence. Medical devices, aviation, railway signaling and some financial systems must show auditors that every requirement is traced to a design and a test. That favors the V-model or a plan-driven process with strong documentation, sometimes with agile iterations inside it.

Risk and uncertainty. If there are major technical or business unknowns, you want a risk-driven approach that learns cheaply before committing, as in the spiral model.

Customer availability. Agile processes assume a customer or product owner who can answer questions and give feedback every week or two. If the customer will only appear at the end (or is a faceless market), you must substitute research and early releases.

Team size, experience and location. A small, experienced, co-located team can run lightly. A large team spread across time zones needs more written agreements. A team new to the domain needs more guidance and shorter loops.

Nature of the work. Building a new product has a start and an end. Operating a service with a stream of unpredictable small requests (bug fixes, support tickets) is continuous flow, which suits Kanban better than fixed-length iterations.

Dials that shape the choice

Volatility How often will requirements change?
Size How big, how costly to get wrong?
Regulation Must you prove traceability?
Uncertainty How many big unknowns remain?
1 / 2

A decision table

The table below maps typical situations to a sensible starting point. Read it as "begin here, then adapt", not as a law.

Situation Good starting point Why
Small project, requirements fixed and well understood Waterfall or simple incremental Little uncertainty, so a plan costs little
Safety-critical or regulated, requirements stable V-model Traceability from requirement to test
Regulated, but features still evolve V-model with iterations, or agile with strict documentation Evidence plus feedback
Large, novel, high technical risk Spiral Risks are retired first, with exit points
New product, volatile requirements, customer available Scrum (iterations of 1 to 4 weeks) Fast feedback and replanning
Maintenance, support or operations work Kanban Continuous flow, limit work in progress
Fixed contract, fixed scope, short deadline Incremental with a firm plan Delivers the highest-value parts first

Notice how often the answer is "X with some Y". Real organizations mix models: a bank might run a spiral-style discovery for a new payment platform, then Scrum for the build, and a V-model-style test and evidence phase for the regulator.

Hybrids are normal

Water-Scrum-Fall is a half-joking name for a very common pattern: requirements and budgets are set up front in the old way, development runs in sprints, and release is a formal, gated handover. It is not pretty, but it is what many large organizations can actually adopt.

A cleaner hybrid is plan the big picture, adapt the details: decide the goals, the architecture direction and the major milestones early, but let the order and detail of features change every iteration. This follows the same logic as estimation and planning in agile teams: the further away something is, the less detail you plan.

Common mistakes

  • Choosing by fashion. "Everyone does agile" is not an analysis. Agile is excellent when learning and change dominate; it does not remove the need to know what a regulator demands.
  • Copying a framework's ceremonies without its purpose. Daily meetings and sprint reviews are only useful if the feedback they create is acted upon.
  • Ignoring the organization. A process that needs a weekly customer cannot succeed if the customer sends one email a quarter.
  • Never revisiting the choice. A project's dials move. The startup that begins with volatile requirements may later need stricter release control when thousands of customers depend on it.

Review the process at milestones or retrospectives and ask: "Is this process still giving us fast feedback and acceptable risk?" If not, change something small, then measure again.

A recommender stand-in

The code below scores each model against six yes/no or low/high answers. Each answer adds or removes points from the models it affects, using the rules of thumb above. You can change the project object and rerun to see how the recommendation moves. The scoring is deliberately simple so that you can read every rule; it will not replace thinking.

JS Recommend a process model

Press Run (or Ctrl + Enter) to see the console output here.

No output.

The three sample projects land on three different answers, which matches intuition: Scrum for the consumer app, the V-model for the regulated firmware, and Kanban for the support stream. Try making the firmware project volatility: "high" and watch the V-model lose ground; that is the situation where teams reach for a hybrid.

Turning a recommendation into a process

A model name is only a starting point. Write down, in a page or less, how your team will actually work:

  1. How work is planned and who decides priorities.
  2. How long a cycle is and what "done" means for one item (Definition of Done).
  3. How and when you test (testing levels).
  4. How changes to scope are handled (change management).
  5. When you will review whether the process is working.

Small, explicit agreements beat a thick process manual that nobody reads.

A second look: the cost of being wrong

Sometimes it helps to ask what happens if you pick badly. If you choose waterfall and requirements move, you pay a large rework bill late. If you choose Scrum on a regulated project and neglect traceability, you pay at audit time. If you choose a spiral for a tiny project, you pay a modest overhead. The asymmetry is useful: when you are unsure, prefer the process whose failure mode is cheapest and most visible. Short iterations with early releases usually fail loudly and early, while long plans tend to fail quietly and late.

Check your understanding

Quick check ยท 5 questions

of 5 answered You scored / 5 ยท perfect!

  1. 1. A team is building firmware for a medical device with stable, regulated requirements. Which starting point fits best?

    Regulators ask for evidence that each requirement is designed and tested. The V-model builds that traceability in.
  2. 2. Which situation points most strongly toward Kanban?

    Kanban manages continuous flow and limits work in progress, which suits a stream of unplanned small items.
  3. 3. Why do many organizations use hybrid processes?

    A project may need risk-driven discovery, iterative building and a formal release gate. Mixing models fits those different needs.
  4. 4. Which is a mistake when adopting a process?

    Agile ceremonies exist to create and use feedback. Without a customer to supply it, they become empty rituals.
  5. 5. When you are unsure which model to use, what is a sensible tiebreaker?

    Short feedback cycles tend to reveal problems early and loudly, while long plans hide them until late.

Summary

  • Choose a process from project characteristics: requirement volatility, size, regulation, risk, customer availability, team and the nature of the work.
  • Plan-driven models suit stable or regulated work; spiral suits large, risky, novel work; Scrum suits volatile requirements with an available customer; Kanban suits continuous flow.
  • Hybrids are normal, and the choice should be revisited as the project changes.
  • Write a short, explicit description of how your team will work, and review it regularly.

Next: Gathering Requirements.