i

Description

Une collection d'outils numérotés, disposés côte à côte pour comparaison

Source

State Government Photographer · original page

Licence

CC0 Public domain

Original photo, unretouched. The background is a version of it stylised by the site.

Comparison of project management methods

You have two /5 rating grids (general software project, then data/BI project) across nine aspects, for twelve approaches: waterfall, V-model, spiral/RUP, PRINCE2/PMBOK, DSDM, Scrum…

In short

• You have two /5 rating grids (general software project, then data/BI project) across nine aspects, for twelve approaches: waterfall, V-model, spiral/RUP, PRINCE2/PMBOK, DSDM, Scrum, Kanban, Scrumban, XP, Lean startup, Shape Up, SAFe.
• Each grid is a table heatmap (one colored square per rating) to spot weaknesses and strengths at a glance.
• You will learn to read these ratings with caution: they are expert, contextual assessments that collapse if the context changes.
• You will see how to weight by project profile, and which contexts tip the decision.
• You are not looking for “the best method” but the least bad one for your context: the decision is made in Choosing Your Method and Hybridizing.

On this page

1. 1. Warning: what these ratings are (and are not)
2. 2. The nine defined aspects
3. 3. Grid 1 — general software project
4. 4. Grid 2 — data / BI project
5. 5. What changes between the two grids
6. 6. Weighting by project profile
7. 7. Critical reading of the ratings
8. 8. Contexts That Shift the Balance
9. 9. AI Contribution
10. 10. See also

Level: 🟢→🔴

1. Warning: what these ratings are (and are not)#

🟢 Novice — A 4/5 rating does not mean “the method is 80% good.” It means: in the stated context, on this aspect, an experienced practitioner expects a result clearly above the average of the methods compared. The principles of rating (scale, context, Goodhart's law) are described in RATING.

Shared context for the ratings: team of 4 to 12 people, project of 6 to 18 months, mid-sized organization, competent practitioners (a poorly applied method has no rating: see Pitfalls and good practices in project management (cross-functional summary)).

2. The nine defined aspects#

🟡 Practical — The ratings are consistent with the other guides: same vocabulary as Predictive Methods (waterfall, V-model, spiral, RUP, PMBOK, PRINCE2, Merise, CMMI) and Agile and Lean Methods (Manifesto, Kanban, XP, Lean, Shape Up, DSDM, SAFe, DevOps…).

Note on aspect 5: in grid 1 it measures fit for a software product; in grid 2, fit for a data/BI project (source quality, model, acceptance testing on numbers, historization, kit-based delivery).

3. Grid 1 — general software project#

Heatmap legend: 🟥 1 · 🟧 2 · 🟨 3 · 🟩 4 · 🟦 5. Columns: Predictability · Adaptation · Risk · Lightweightness · Fit for software · Adoption · Scale · Governance · Quality.

Justification for the most discriminating ratings (one sentence each):

• Waterfall, adaptation 1/5: real feedback arrives only at the end of the project, so any late change is expensive.
• V-model, quality 3/5: verification is planned early, but executed late and in one block.
• Spiral, risk 5/5: the cycle is entirely driven by risk reduction, at the cost of difficult adoption 2/5.
• PRINCE2, governance 5/5: tolerances, stages, and committee provide a complete audit trail; lightweightness 2/5 as a tradeoff.
• Scrum, fit 5/5: designed for a software product with evolving needs; predictability 3/5 because forecasting is probabilistic (velocity), not committed to a long-term scope.
• Kanban, lightweightness 5/5, risk 2/5: no ceremonies, but no built-in risk treatment mechanism.
• XP, quality 5/5 but scale 2/5: technical practices are the most demanding, multi-team coordination is missing.
• Lean startup, predictability 1/5: by design, you do not commit to an outcome, you commit to experiments.
• Shape Up, scale 2/5: designed for one product and small autonomous teams.
• SAFe, lightweightness 1/5: large-scale coordination has a high organizational cost; scale 5/5 in return.

4. Grid 2 — data / BI project#

Additional context: a SQL Server datamart fed by operational sources (field sales force application, for example), Power BI reports, acceptance testing on figures by the business, delivery in versioned installation kits. Aspect 5 here measures fit for a data/BI project (source profiling, historization, replay, migrations).

Justification for the gaps versus grid 1 (what changes when it's data):

• Kanban rises (fit 5/5): data workload is made up of requests with heterogeneous size and urgency (incident, enhancement, figure correction); flow and classes of service are a natural fit.
• Scrum drops (fit 4/5, predictability 3/5): the “potentially shippable” increment is harder to define when data depends on heavy environments and business acceptance testing; sprint slicing works poorly when a source imposes its own schedule. Details in Scrum and data.
• XP drops (leanness 2/5, fit 3/5): testing, continuous integration, and pairing practices exist but require data testing tooling that is less mature than application code tooling.
• Waterfall / V drop in predictability: the plan assumes the sources are known; profiling often reveals surprises. But for a like-for-like migration, V goes back up (predictability 4/5 in this specific case).
• SAFe drops (fit 2/5): few data projects reach the size where a release train is justified.
• Lean startup stays high in adaptability: report prototyping and exploratory analysis are true MVPs.

5. What changes between the two grids#

🟡 Practical — The eight traits of data projects that shift ratings:

Details and countermeasures in Methods Specific to Data Projects and in the matrices Maturity Matrix — Data Project Management (PD) and Maturity Matrix — Development of a BI Tool / Application.

6. Weighting by project profile#

🟡 Practical — The unweighted average assumes the nine aspects all matter equally: that's false in every real case. You choose weights (0 to 3) based on context. Two profiles, applied to grid 2 (data) — the weights are illustrative choices, not truths:

Results (weighted score = Σ weight × score / Σ weights), for five approaches:

Reading: profile A brings V and PRINCE2 up; profile B raises Kanban and Lean startup. But note two warning signals.

7. Critical reading of the ratings#

🔴 Expert — Five reasons not to treat ratings as measurements:

1. No single empirical measurement. Comparative studies between methods are rare, heterogeneous, and context-sensitive; we don't have robust data that would allow us to quantify these ratings. They are expert judgments with supporting arguments, so they are debatable and revisable.
2. Within-method variance exceeds between-method variance. Two teams doing “Scrum” differ more from each other than Scrum differs from well-executed Kanban. Quality of implementation (coaching, management commitment, technical discipline) matters more than the method name.
3. Methods are often mixed. Almost no real project is “pure”: see the next article.
4. Halo effect of criteria. “Leanness” and “adoption” are correlated in evaluators' minds; the average counts them twice. Any aggregate is debatable.
5. The organization is the hidden variable. A “scaling” score says nothing about management, architecture, or contracts, which determine the outcome.

Useful correlations to remember (qualitative):

• Predictability and adaptation are in tension: the more you lock things down, the less you adapt. Methods that score high on both (DSDM, Shape Up) do so by fixing timeline and budget while varying scope.
• Leanness and governance are in tension: strong governance requires artifacts.
• Deliverable quality depends more on engineering practices (tests, review, continuous integration) than on the planning method: XP enforces them explicitly, but you can add them to any method.
• Scaling always costs leanness: each coordination layer adds meetings. The best way to scale is reducing dependencies.

8. Contexts That Shift the Balance#

🟡 Practical — A single context change can overturn the choice. Here are the most common shifts to revisit at each project milestone:

Trade-offs are laid out step by step in Choosing Your Method and Hybridizing.

9. AI Contribution#

🤖 AI — Useful for exploring scenarios and clarifying criteria, risky for deciding.

Prompt example: “Here is my grid (table) and my constraints (fixed-price contract, 6 people, biweekly business acceptance testing). Vary the weights by ±1 on each aspect and tell me which approaches change rank.”

What AI does not replace: knowledge of people, the actual contract, and your organization’s risk tolerance. See also AI in Service of Project Management (General Then Data).

10. See also#

• Before: Predictive Methods (waterfall, V-model, spiral, RUP, PMBOK, PRINCE2, Merise, CMMI) · Agile and Lean Methods (Manifesto, Kanban, XP, Lean, Shape Up, DSDM, SAFe, DevOps…) · Methods Specific to Data Projects.
• After: Choosing Your Method and Hybridizing · CI/CD · Pitfalls and good practices in project management (cross-functional summary).
• Framework: Scrum · Scrum and data · Fundamentals.
• Tools: comparison of tools · planning tools.
• Maturity: Governance and Steering (GOU) · Continuous Improvement and Learning (AMC) · Delivery and Production Release (LIV).
• Glossary · Rating.

✻ Written by Claude Sonnet 5.5