0.4Methods overview: predictive vs agile
The essentials in 30 seconds
- Predictive: plan everything upfront then execute in sequence — ideal when the need is stable and known.
- Agile: move in short iterations and adapt — ideal when the need is uncertain or evolving.
- Contextual choice: no chapels — choose (or blend) by need, uncertainty and culture.
Running mini-case — Bilal chooses without following fashion
Bilal has two projects: migrating the host (planned cutover, frozen technical steps) and building a client app whose need stays fuzzy. For the migration, he plans everything predictively. For the app, he proposes two-week iterations with client demos. And for the site redesign, he blends: overall scoping then iterative delivery.
Last episode of cases 0.1-0.3 — Bilal fits the method to the need.
1. Two big families
There is no single way to run a project. The predictive approach (V-cycle, waterfall) plans everything in detail upfront then executes sequentially: ideal when the need is stable and well known (construction, industry).
The agile approach moves in short iterations, adapting as it goes: ideal when the need is uncertain or evolving (software, innovation).
2. A contextual choice
The choice is not ideological but contextual: it depends on need stability, uncertainty level and the organization's culture.
Many projects adopt a hybrid approach (predictive scoping + agile delivery). Levels 2–3 deepen predictive, level 4 agility.
🔎 Concrete example — 🔎 Concrete example — Stable or uncertain need?
The choice criterion.
Standardized construction, frozen need → rather PREDICTIVE.
Innovative software, client discovering as they go → rather AGILE.
Big mixed project → HYBRID (scope globally, deliver in iterations).Outcome: a fitting approach, not a followed fashion.
Reflex: first assess need stability.
Section vocabulary
- Predictive approach
- Detailed upfront planning then sequential execution (waterfall, V-cycle). Suited to stable needs.
Simple example: Bilal plans the host migration step by step before the cutover.
What it is not: moving in iterations while discovering the need (that's agile).
Reflex: frozen need and constrained sequence → lean predictive.
- Waterfall / V-cycle
- Predictive models: chained phases (the V adds symmetry between specifications and tests/validations).
Simple example: Specify the site, build it, then test each requirement one by one.
What it is not: delivering in small increments adjusted each cycle.
Reflex: the V = each specification matches a test.
- Agile approach
- Iterative, incremental, adaptive development in short cycles. Suited to uncertain or evolving needs.
Simple example: Bilal delivers the app in demos every two weeks and adjusts.
What it is not: freezing everything upfront then never changing anything.
Reflex: fuzzy or evolving need → iterate and show early.
- Iterative / incremental
- Moving in small cycles (iterations) delivering pieces of value (increments), with regular adjustments.
Simple example: Each Bilal iteration adds a usable piece of the app.
What it is not: waiting until the end to deliver everything as one block.
Reflex: each cycle must produce something showable.
- Hybrid
- Combining both approaches (e.g. predictive scoping + agile delivery), per project needs.
Simple example: Bilal scopes the site globally, then delivers in iterations.
What it is not: applying one single method on principle, everywhere.
Reflex: blend by phase: scope globally, iterate delivery.
In practice — In practice — Choosing the right approach
- For a given project, assess need stability: is it clear and frozen, or uncertain and likely to evolve?
- If the need is stable and the deliverable well defined (construction, standardized rollout) → lean predictive.
- If the need is fuzzy, innovative or evolving, with a client discovering as they go → lean agile.
- Consider hybrid: scope globally (predictive) then deliver in iterations (agile) depending on context.
Ready-to-say script — justifying the method choice (30s)
"Our need is stable and the sequence constrained: we plan everything predictively. There, the need is fuzzy and will evolve: we move in two-week iterations with demos. For this mixed project, we blend: overall scoping, then iterations."
Say to client or owner: tie each choice to need stability, not preference.
Copy-paste template — predictive / agile choice grid
PROJECT: _______________
NEED (stable / uncertain / mixed): _______________
UNCERTAINTY (low / high): _______________
CHOSEN APPROACH (predictive / agile / hybrid): _______________
WHY (1 sentence): _______________
FIRST MILESTONE OR FIRST ITERATION: _______________Small business / tight budget
No heavy ceremony: one iteration = showing something to the client every week, even informally.
Algerian SME
Bilal sets iterations around the client's real availability (short demos, written feedback) to keep the agile pace.
Construction / regulated project
Frozen need and strict standards: embraced predictive, with agile points limited to design.
Startup / innovative product
Need being discovered: short iterations, early user tests — predictive only serving overall scoping.
✓ Best practices
- Assess need stability before choosing the approach.
- Don't oppose methods: combine them if the project suits it.
- Consider the organization's culture.
✗ Common pitfalls
- Choosing a method by fashion rather than context.
- Believing agile = no planning.
- Wanting "pure" agile on a sequential, constrained project.
Key takeaways
- Predictive (waterfall, V-cycle): plan everything upfront. For stable needs.
- Agile: iterate and adapt. For uncertain or evolving needs.
- The choice is contextual, not ideological.
- Hybrid often combines both across project phases.
48h Challenge — apply tomorrow
Take 1 real project, judge its need stability and choose predictive, agile or hybrid (template above). Announce the first milestone or first iteration.
- I can say when predictive fits (stable need).
- I can say when agile fits (uncertain need).
- I've chosen 1 approach for a real project, no fashion-following.
Frequently asked questions
Is agile "better" than predictive?
No: each excels in its context. Agile shines on uncertainty and evolution (software, innovation); predictive stays relevant when the need is stable and the sequence constrained (construction, regulated projects). Opposing the two is sterile: the right reflex is to choose — or combine — by project. Maturity means knowing both and picking what serves the project, not defending a chapel.
Can you do agile on a construction project?
Hardly identically: you don't pour foundations "in iterations" redone every sprint. Some agile principles (collaboration, regular check-ins, adjustments) stay useful, but the sequential, costly-to-change nature of building calls for a largely predictive approach. The physical context imposes its constraints. The design or interior-fitting phase, however, can sometimes include more agile loops.
When to prefer an agile approach?