# sfeir-sdlc-pdlc-articulation-2026-07-22

## Veille

SFEIR analysis (consulting-firm voice, "an engineer's reading") articulating two frameworks too often conflated: the **SDLC** (Software Development Life Cycle — *building the software correctly and reliably*) and the **PDLC** (Product Development Life Cycle — *building the right product and succeeding in the market*). Central thesis: the two cycles are not competitors but **nested** — the SDLC is the subset of the PDLC **housed under its development phase**; when a product team reaches the "build" stage, a full SDLC cycle (design → build → test → review → deployment) runs inside it. The SDLC is standardized (**ISO/IEC/IEEE 12207**, 2017 and 2026 editions), with its lineage of models (Waterfall 1970, V-model, iterative/spiral, **Agile 2001**, **DevOps/DevSecOps 2009+**) and its **DORA** metrics (throughput, stability, MTTR, change failure rate). The PDLC, being the umbrella cycle, runs from **ideation/discovery** to **market withdrawal** (not to be confused with the marketing **PLC** of Theodore Levitt, 1965, which describes a *commercial curve*, not *organized work*: "the PLC observes a curve; the PDLC organizes work"). **Tipping point**: the SDLC natively addresses **only one risk in four** — via **Marty Cagan's "Four Big Risks"** framework (Value → PM, Usability → Designer, Feasibility → Lead Engineer, Business viability → PM) — an organization excellent at SDLC but blind to PDLC produces "software nobody wants" — John Cutler's **"feature factory"** (success measured by output, not outcome). **Why AI changes everything**: generative AI **compresses the SDLC** (Google/JetBrains data, May 2026: **~85% of developers** regularly use coding agents, **~41% of new code** is AI-generated; implementation goes from weeks to hours), so the **bottleneck shifts upstream** — deciding *what* to build (Marty Cagan, April 2026: "when the cost of delivery collapses, the bottleneck shifts to discovery"). Consequences: DORA 2025 (~5,000 professionals, 90% AI adoption) shows a **positive correlation with throughput but a negative one with stability** (more unvalidated features means instability and rework); Andrew Ng (AI Startup School, July 2025) reports teams **reversing the "1 PM for 4 engineers" ratio to "2 PMs for 1 engineer"**; and with **spec-driven development**, the PDLC/SDLC boundary becomes **porous** (the product spec becomes directly executable by agents). **What a CIO should take away**: an augmented SDLC becomes a **market standard, not a differentiator** — the junction with the product must be instrumented, **executable specifications** demanded as input, technical metrics cross-referenced with outcome metrics, and the role of "feature supplier" **refused**. For a CPO: the shift of the bottleneck toward discovery is both a **promotion** (product judgment becomes scarce again) and a **notice to act** (industrialize discovery to reach parity with the SDLC). SFEIR's in-house framework ("Designing and building in the agentic era" — **11-phase cycle** + **Software Factory 10x**) is positioned as the answer on the engineering side, with the **articulation of the two cycles** as the next lever. Conclusion: "as code becomes a commodity, margin shifts toward product judgment and governance."

## Titre Article

SDLC vs PDLC : quelle différence, et pourquoi l'IA change tout

## Date

2026-07-22

## URL

https://www.sfeir.com/articles/sdlc-vs-pdlc-difference-articulation/

## Keywords

SDLC, Software Development Life Cycle, PDLC, Product Development Life Cycle, software life cycle, product life cycle, PLC, product life cycle, Theodore Levitt, ISO/IEC/IEEE 12207, 12207, Waterfall, V-model, iterative, spiral, Agile, DevOps, DevSecOps, CI/CD, DORA, DORA metrics, throughput, stability, MTTR, change failure rate, lead time, velocity, discovery, ideation, product-market fit, adoption, retention, NPS, Marty Cagan, Four Big Risks, four major risks, value, usability, feasibility, business viability, Product Manager, CPO, product lead, Lead Engineer, designer, John Cutler, feature factory, output vs outcome, bottleneck, bottleneck shift, upstream, what to build, cost of delivery, generative AI, coding agents, 85% of developers, 41% of code AI-generated, Google JetBrains, vibe coding, The New SDLC With Vibe Coding, Andrew Ng, PM-to-engineer ratio, 2 PMs per engineer, spec-driven development, executable specification, PDLC SDLC porosity, porous boundary, CIO, CIO, differentiator, market standard, product-engineering junction, end-to-end governance, SFEIR agentic framework, 11-phase cycle, Software Factory 10x, augmented PM, instrumented discovery, agent-based prototyping, nesting of cycles, subset, product judgment, code commodity

## Authors

SFEIR (voix éditoriale du cabinet)

## Ton

**Profile**: pedagogical-strategic thought-leadership piece from a consulting firm (SFEIR), addressed to CIOs, CPOs, and technical leadership. Canonical structure "vocabulary clarification → articulation → AI impact → role-based recommendations → FAQ," with a **comparison table** dimension by dimension (scope, purpose, question, actors, metrics, horizon, risks) and an FAQ section resolving frequent confusions ("does the PDLC replace the SDLC?", "PDLC vs PLC?").

**Style**: didactic and structured — first lays out standardized definitions (referencing ISO/IEC/IEEE 12207, historical models), then draws on recognized authorities (Marty Cagan for the Four Big Risks and the bottleneck shift, John Cutler for the feature factory, Andrew Ng for the PM/engineer ratio reversal, DORA for stability data). Recurring **hammer phrases**: "the PLC observes a curve; the PDLC organizes work"; "the SDLC natively addresses only one risk in four"; "as code becomes a commodity, margin shifts toward product judgment and governance." Distinguishes **correlation from causation** (DORA 2025 data: "correlations, not causation"). Closes on a deliberate **product positioning** of SFEIR's in-house framework (11-phase cycle, Software Factory 10x), presented as the already-delivered building block to which the articulation of the two cycles must now be added.

## Pense-betes

- **The distinction in one sentence.** **SDLC** = "building the software **correctly and reliably**" (question: *how to ship it?*); **PDLC** = "building the **right** product, succeeding in the market" (question: *what to build, and why?*). The SDLC is a **subset** of the PDLC, not its competitor.
- **Nesting, not substitution.** The SDLC is **housed under the development phase** of the PDLC. When the product team reaches the "build" stage, a full SDLC cycle (design → build → test → review → deployment) runs inside it. Direct answer to the FAQ: "the PDLC does not replace the SDLC, the two are nested."
- **Do not confuse PDLC and PLC.** The **PLC** (product life cycle, Theodore Levitt, HBR 1965 — introduction/growth/maturity/decline) describes a **commercial trajectory** that guides marketing. The **PDLC** describes the **design and building process**. Formula: "the PLC observes a **curve**; the PDLC organizes **work**."
- **The SDLC covers only one risk in four.** Via **Marty Cagan's "Four Big Risks"**: **Value** (PM — "will they come, will they choose it?"), **Usability** (Designer — "will they know how to use it?"), **Feasibility** (Lead Engineer — "can it be built?"), **Business viability** (PM — "does it work for the business?"). The SDLC natively addresses only **technical feasibility**. SDLC excellence + PDLC blindness = "software nobody wants" = **John Cutler's** **feature factory** (success measured by **output**, not **outcome**).
- **Why AI changes everything: the compression of the SDLC.** **Google/JetBrains data, May 2026**: **~85% of developers** regularly use coding agents; **~41% of new code** is AI-generated. Implementation goes from **weeks to hours** — but requirements, architecture, and verification remain at **human pace**.
- **The bottleneck shifts upstream.** **Marty Cagan (April 2026)**: "when the cost of delivery collapses, the bottleneck shifts to discovery — deciding **what** to build." What is scarce is no longer code, it is **product judgment**.
- **The DORA 2025 signal (read it carefully).** Google Cloud survey (~5,000 professionals, 90% AI adoption): correlation **positive with delivery throughput**, but **negative with stability** (correlations, **not causation**). Interpretation: shipping **unvalidated features** faster creates instability and rework — exactly what PDLC discipline is meant to prevent upstream.
- **The ratio reversal.** **Andrew Ng (AI Startup School, July 2025)**: some teams propose reversing the historical **"1 PM for 4 engineers"** ratio to as much as **"2 PMs for 1 engineer"** — depending on context. A sign that the center of gravity of effort is moving back toward product definition.
- **The boundary becomes porous.** With **spec-driven development**, the PDLC's design artifact **feeds directly** into the SDLC: "the boundary between the two cycles becomes porous." The **product specification becomes executable** by agents.
- **What a CIO must acknowledge.** Optimizing the SDLC alone is no longer enough: an augmented SDLC becomes a **market standard, not a differentiator**. Actions: (1) **instrument the junction** with the product; (2) **demand executable specifications** as input; (3) **cross-reference** technical metrics with **outcome** metrics; (4) **refuse** the role of "feature supplier." Named risk: an **artisanal PDLC facing an industrialized SDLC** creates an "untenable imbalance."
- **What a CPO must acknowledge.** The bottleneck shift is both a **promotion** (product judgment becomes scarce again) **and** a notice to act: **equip discovery** (agent-based prototyping, accelerated validation of the four risks) to reach **industrialization parity** with the SDLC. "The CPO now holds the company's critical path."
- **SFEIR's in-house positioning.** The "Designing and building in the agentic era" framework — **11-phase cycle** + **Software Factory 10x** — settles the engineering side; the **next lever** is the **articulation of the two cycles** (augmented PM, PDLC/SDLC porosity, end-to-end governance). Conclusion: "as code becomes a **commodity**, margin shifts toward **product judgment and governance**."
- **Related**: **SDLC / ADLC / agentic cycle** cluster (BMAD-Method agentic-AI urbanism 2026-02-04; The New SDLC With Vibe Coding, Google May 2026; SFEIR "architect in the AI era" 2026-07-15; SFEIR 11-phase cycle); **DORA** (DORA 2026-04-21 ROI/J-curve, State of AI-assisted Software Development 2025); **feature factory / output vs outcome** (John Cutler); **spec-driven / Software Factory** (StrongDM software factory 2026-02-06; AI spec-driven approach); **KDLC** (Ashish Singh, knowledge life cycle, June 28, 2026) and **compounding knowledge lifecycle** (Klaassen 2026-07-02) as neighboring life-cycle frameworks.

## RésuméDe400mots

SFEIR clarifies two frameworks often conflated. The **SDLC** (Software Development Life Cycle), standardized by **ISO/IEC/IEEE 12207** (2017, 2026), structures **software production** — requirements gathering, design, development, testing/QA, deployment, maintenance — with its lineage of models (Waterfall 1970, V-model, iterative/spiral, **Agile** 2001, **DevOps/DevSecOps** 2009+) and its **DORA** metrics (throughput, stability, MTTR, change failure rate). Its purpose: "building the software **correctly and reliably**." The **PDLC** (Product Development Life Cycle) is the **umbrella cycle**: from ideation/discovery to market withdrawal, it aims to "build the **right** product." Not to be confused with Theodore Levitt's **PLC** (1965), which describes a **commercial curve**; "the PLC observes a curve, the PDLC organizes work."

**Articulation**: the cycles are **nested** — the SDLC is the subset of the PDLC housed under its **development phase**. Critical point via **Marty Cagan's "Four Big Risks"** (Value, Usability, Feasibility, Business viability): the SDLC natively addresses only **technical feasibility** — "one risk in four." An organization strong in SDLC but blind to PDLC becomes **John Cutler's** **"feature factory,"** which measures success by **output** rather than **outcome**.

**Why AI changes everything**: generative AI **compresses the SDLC** (Google/JetBrains, May 2026: **~85%** of developers use coding agents, **~41%** of new code is AI-generated; implementation goes from weeks to hours). The **bottleneck shifts upstream** — deciding *what* to build (**Cagan**, April 2026). Three consequences: **DORA 2025** (~5,000 professionals, 90% adoption) shows a correlation that is **positive with throughput but negative with stability** (correlations, not causation) — more unvalidated features, more rework; **Andrew Ng** (July 2025) reports the reversal of the ratio **"1 PM / 4 engineers" to "2 PMs / 1 engineer"**; and **spec-driven development** makes the **PDLC/SDLC boundary porous** (the spec becomes executable by agents).

**Recommendations.** For the **CIO**: an augmented SDLC is now a **market standard, not a differentiator** — instrument the product junction, demand **executable specifications**, cross-reference technical and outcome metrics, refuse the role of "feature supplier"; an artisanal PDLC facing an industrialized SDLC is an "untenable imbalance." For the **CPO**: both a promotion **and** a notice to act — **equip discovery** to reach industrialization parity. SFEIR positions its in-house framework (**11-phase cycle** + **Software Factory 10x**) as the answer on the engineering side, with the **articulation of the two cycles** as the next lever. Conclusion: "as code becomes a commodity, margin shifts toward product judgment and governance."

## GrapheDeConnaissance

- SDLC —fait_partie_de→ PDLC (METHODOLOGIE, 0.95)
- SDLC —est_instance_de→ ISO/IEC/IEEE 12207 (DOCUMENT, 0.9)
- PDLC —s_oppose_à→ PLC (product life cycle, Theodore Levitt 1965) : le PLC observe une courbe commerciale, le PDLC organise un travail de conception (AFFIRMATION, 0.85)
- SDLC —affirme_que→ le SDLC ne traite nativement qu'un risque sur quatre (la faisabilité technique) parmi les Four Big Risks de Cagan (AFFIRMATION, 0.9)
- Marty Cagan —a_créé→ Four Big Risks (CONCEPT, 0.92)
- Four Big Risks —s_applique_à→ répartition des responsabilités produit : Valeur/Viabilité (PM), Utilisabilité (Designer), Faisabilité (Lead Engineer) (AFFIRMATION, 0.9)
- John Cutler —a_créé→ feature factory (CONCEPT, 0.9)
- feature factory —observé_dans→ organisations fortes en SDLC mais aveugles au PDLC, qui mesurent le succès à l'output plutôt qu'à l'outcome (AFFIRMATION, 0.88)
- IA générative —réduit→ le coût et la durée du SDLC : implémentation de semaines à heures (AFFIRMATION, 0.9)
- Google JetBrains —mesure→ ~85 % des développeurs utilisent régulièrement des agents de code et ~41 % du nouveau code est généré par IA (mai 2026) (MESURE, 0.9)
- Marty Cagan —affirme_que→ quand le coût du delivery s'effondre, le goulot d'étranglement se déplace vers l'amont : décider quoi construire (avril 2026) (AFFIRMATION, 0.92)
- Rapport DORA 2025 —mesure→ adoption IA à 90 % corrélée positivement au débit de livraison mais négativement à la stabilité (corrélations, non causalités) (MESURE, 0.88)
- Andrew Ng —affirme_que→ certaines équipes proposent d'inverser le ratio historique de 1 PM pour 4 ingénieurs jusqu'à 2 PM par ingénieur (juil. 2025) (AFFIRMATION, 0.85)
- approche spec-driven —permet→ rendre poreuse la frontière PDLC/SDLC : la spécification produit devient directement exécutable par des agents (AFFIRMATION, 0.85)
- SFEIR —recommande→ une DSI doit instrumenter la jonction produit, exiger des spécifications exécutables et refuser le rôle de fournisseur de features (AFFIRMATION, 0.88)
- SFEIR —affirme_que→ à mesure que le code devient une commodité, la marge se déplace vers le jugement produit et la gouvernance (AFFIRMATION, 0.9)
- cycle SFEIR à 11 phases —résout→ le versant ingénierie (SDLC augmenté) ; le levier suivant est l'articulation SDLC/PDLC (AFFIRMATION, 0.82)

---
Canonical: https://www.thekb.eu/en/fiches/sfeir-sdlc-pdlc-articulation-2026-07-22/