lifecycle-play-advisor
À propos
Cette compétence diagnostique le stade du cycle de vie d'un produit pour recommander s'il faut le prolonger, le remplacer ou le retirer. Elle guide les utilisateurs à travers un triage pour identifier la cause profonde du déclin et teste d'abord les options de prolongation. Utilisez-la pour trancher les débats sur un produit en perte de vitesse avec un diagnostic structuré plutôt que par la spéculation.
Installation rapide
Claude Code
Recommandénpx skills add deanpeters/Product-Manager-Skills -a claude-code/plugin add https://github.com/deanpeters/Product-Manager-Skillsgit clone https://github.com/deanpeters/Product-Manager-Skills.git ~/.claude/skills/lifecycle-play-advisorCopiez et collez cette commande dans Claude Code pour installer cette compétence
Documentation
Lifecycle Play Advisor
Purpose
Work out what to do with a product that has stopped growing. Three plays are available — extend, replace, retire — and this skill gets you to the right one through diagnosis rather than debate, then hands you to whichever skill executes it.
Most teams argue the play before establishing the stage. The argument is unwinnable that way, because nobody has agreed on what's actually happening to the product. Four questions fixes that.
This skill is deliberately willing to say "nothing yet." A mature product throwing off margin with manageable support cost doesn't need a play; it needs to be left alone and watched.
Input
Works best with: The product, and what makes you think something needs to change.
Also useful: Revenue trend and over how long, support load, what customers say, whether an internal driver (cost, capacity, strategy) is really behind the question, and what investment appetite exists.
Anything supplied with the invocation itself — text after the skill name, a pasted context dump, or
an appended ARGUMENTS: line — counts as answers already given. Use it and skip whatever it
covers; don't re-ask.
Arriving empty-handed? That works too. The diagnosis runs on questions a PM can answer from what they already know — no report-pulling required. "I don't know" is a valid answer and becomes a labeled assumption in the recommendation.
Example invocations:
Our reporting module has been flat six quarters and people want to rebuild it. Extend, replace, or retire?Leadership wants to kill our parts module. Walk me through whether that's right.
Key Concepts
The Three Plays
- Extend — add a variant or capability to the existing line. The product keeps running; you add alongside. Cheapest, lowest risk, most often skipped because it isn't exciting.
- Replace — ship a successor and phase the old one out. GTM and EOL run simultaneously, for two products that compete with each other. The expensive play.
- Retire — phase out with no successor of your own. Customers land elsewhere, possibly with a competitor. The goal becomes losing the product without losing the customer.
And the fourth answer that isn't a play: harvest — stop investing, keep running, set a review date.
Diagnosis Before Play
The seven transition questions establish stage; the pressure source discriminates the plays. Run both before recommending anything, and show your work — a recommendation the user can audit is one they can defend to someone else.
The Extension Bias
Default toward the cheapest play that addresses the actual pressure. Extension is under-considered in almost every room, because replacement is more interesting to build and retirement is more decisive to announce. Test extension explicitly, and record why it failed when it does.
Anti-Patterns (what this is NOT)
- Not a growth strategy tool. Where the next tranche of growth comes from is a different
question — see
ansoff-matrixandorganic-growth-advisor. - Not a kill switch. "Nothing yet" and "harvest" are real outcomes.
- Not the execution. It routes to the skill that runs the chosen play.
Facilitation Source of Truth
Use workshop-facilitation as the interaction protocol. Give
the heads-up at the start — four questions, about fifteen minutes — and let the user dump context
to skip ahead.
Application
This skill asks up to 4 adaptive questions, then recommends a play with its reasoning, its hazards, and a route out.
Question 1 of 4 — What product, and what makes you think something's wrong?
Agent asks:
"Which product are we looking at, and what prompted the question? The prompt matters as much as the product — flat revenue, rising support costs, a competitor move, and an exec remark lead to very different diagnoses."
Listen for whether the trigger is internal or external. An internal trigger — manufacturing wants the line, engineering wants to rewrite, finance wants the cost out — is legitimate, but it means the customer-facing case for change is weak and the transition has to be carried entirely by you. Name that early; it changes what the play costs.
Question 2 of 4 — Where is it actually in the lifecycle?
Agent asks:
"Let's establish the stage. Seven quick reads — yes or no, gut answers are fine:
- Is defending market share still profitable?
- Are loyalty efforts no longer retaining users?
- Are legacy support costs becoming unsustainable?
- Should we be running migration campaigns instead of loyalty ones?
- Are we losing money on certain distribution channels?
- Are our value-add bundles losing effectiveness?
- Is our data shifting from predictive to transitional?"
Score it:
| Yes count | Stage | Implication |
|---|---|---|
| 0-1 | Mature and healthy | No play needed. Invest or leave alone |
| 2-3 | Mature and softening | Extension territory; watch quarterly |
| 4-5 | Crossing into decline | Pick a play deliberately, now |
| 6-7 | In decline | Replace or retire; extension likely just delays |
Say the count back and name the pattern, not just the number. Yeses clustered on 5 and 6 (channel and price) point somewhere very different from yeses on 3 and 7 (support cost and data) — the first pattern is a distribution problem wearing a product costume.
Question 3 of 4 — What's actually driving the pressure?
Agent asks:
"Three sources. Which is loudest?
- Demand-side — customer needs shifted, segments moved, a competitor differentiated
- Supply/cost-side — components, production cost, support load, or a strategy change
- Capability-side — the technology is obsolete or the architecture has hit its limit
Pick one as primary, even if two apply."
Why this discriminates:
| Pressure | Points toward | Because |
|---|---|---|
| Demand-side | Extend | The core still solves a real problem for someone — find them |
| Supply/cost-side | Replace | The problem is your economics, not the customer's need |
| Capability-side | Replace or Retire | Depends on whether the need survives the technology |
Name the trap out loud: supply-side pressure arrives at the roadmap disguised as a customer problem. "We need to rebuild this" often means "our costs are bad." That's a legitimate reason for a replacement — but saying it plainly changes what success looks like and who has to carry the transition.
Question 4 of 4 — Test the extension play
Agent asks:
"Before I recommend anything more expensive, four questions on extending what you have:
- Is there a segment a variant could serve?
- Is there a capability you could add without re-architecting?
- Would a repackage or rebrand reach a different buyer?
- Is the decline in the product, or in the channel or price around it?"
Question 4 catches the most expensive misdiagnosis in the set: a healthy product condemned because the distribution channel stopped working or the bundle lost its edge. Fixing the lever is far cheaper than replacing the product — and the broken lever carries straight over to the successor if you don't.
If all four are no, extension is genuinely off the table, and now the user has the "no, because…" on record for when someone asks in three months.
Final Step: Recommend the play
Deliver this structure:
## Lifecycle Play Recommendation: [Product]
**Stage:** [stage] — [n]/7 transition questions, clustered on [which]
**Primary pressure:** [demand / supply-cost / capability] — [evidence]
**Trigger was:** [internal / external] — [what that means for the transition]
### Recommended Play: [Extend / Replace / Retire / Harvest / Nothing yet]
**Because:** [two or three specifics from their answers]
**Extension test:** [passed and how, or failed and why — always state it]
**What this play costs:** [rough shape]
**What you're accepting by choosing it:** [the tradeoff, plainly]
### Hazards to plan for
- [The 2-3 hazards that apply to this play, from the seven]
### Assumptions I made
- [Anything answered "I don't know"]
Then offer numbered next steps:
"Where next?
- Run the retirement properly —
eol-readiness-advisorfor the go/no-go, oreol-processfor the whole thing - Build the replacement risk register — see
product-lifecycle-plays, especially the seven hazards - Plan the extension —
organic-growth-advisorfor which growth path the variant serves - Run this across the whole line — the portfolio worksheet in
product-lifecycle-plays
Pick a number, combine them, or tell me what you'd rather do."
On a Replace recommendation, always add one line: "A replacement is two funded workstreams — a launch and a retirement. If the retirement isn't funded, you've chosen an extension with extra steps."
Examples
Example A — SaaS, the answer is "your channel is broken, not your product"
PM: "Our Parts module has been flat for five quarters. Leadership wants to fold it into the core platform and stop selling it separately."
Agent (Q1): "What prompted it — a customer signal, or an internal one?"
PM: "Internal. It's the smallest line item and someone noticed."
Agent: "Noted — that's an internal trigger, which means there's no customer pressure to help carry a transition. Let's see whether the product agrees with leadership."
Agent (Q2): runs the seven questions. Yeses on 5 (losing money on certain channels) and 6 (bundles losing effectiveness). Everything else no.
Agent: "2 out of 7 — mature and softening, not declining. And the pattern is the interesting part: both of your yeses are on channel and price. Nothing about the product itself came back negative. Support cost is fine, retention is fine."
Agent (Q3): "What's driving the pressure?"
PM: "Honestly? We sell it through the same motion as everything else and it doesn't fit. Reps don't lead with it."
Agent (Q4): extension test. Segment — yes, distributors have asked for a read-only tier. Capability — yes, a read-only view is minor work. Repackage — yes, that's exactly the ask. Channel or product — channel.
Recommendation: Extend, and fix the channel. Stage says mature-softening. Both decline signals sit on distribution and pricing, not the product. A read-only distributor tier addresses a segment that has actively asked, and it's small work. Folding it into the core would have retired a healthy product to solve a sales-motion problem — and the sales-motion problem would have survived the merge.
What you're accepting: a distributor tier means supporting a new buyer type. Small, but not zero.
Example B — Industrial, the answer is "replace, and fund the retirement"
PM: "Manufacturing wants the NFA-200 line for NFA-500 capacity. So we're retiring the NFA-200."
Agent (Q1): "That's an internal trigger — a capacity constraint, not a customer need. Worth flagging up front: the customer-facing case for change is weak here, so your organization carries the entire transition. What's the revenue trend?"
PM: "Flat four quarters. Service revenue's actually growing."
Agent (Q2): runs the seven. Yeses on 4, 6, 7, and a soft 1. Support costs fine, retention strong, channels profitable.
Agent: "4 out of 7 — mature and softening, not in decline. Your installed base is loyal, under contract, and service margin is healthy. That matters, because it means this isn't a product failing. It's a factory floor you want back."
Agent (Q3): "Primary pressure?"
PM: "Supply side. Manufacturing capacity."
Agent (Q4): extension test. Segment — partially; a modular I/O variant would open bigger jobs. Capability without re-architecture — no, needs a new backplane. Repackage — no, industrial buyers specify on capability. Channel or product — neither.
Agent: "Here's the thing worth saying out loud: the extension your customers would want is the NFA-500. That's a strong signal the replacement play is right — and it raises the question you need to answer before committing. Is the NFA-500 a drop-in for the installed base?"
PM: "...I'd assumed so. I don't actually know."
Recommendation: Replace — with that question as a gate.
Hazards to plan for: internal misalignment (medium), cannibalization (deliberate, low impact), and poor EOL management (high impact) — which is entirely gated on the drop-in question. If the NFA-500 needs different mounting or site work, you don't have a migration path, you have a project, and an EOL date would be a promise you can't keep.
Assumption labeled: drop-in compatibility unverified.
And the line that always goes on a Replace: a replacement is two funded workstreams — a launch and a retirement. Retrofit engineering unfunded means you've chosen an extension with extra steps.
Common Pitfalls
Pitfall 1: Recommending a Play Before the Stage Is Established
Symptom: The user names a play in their first message and the conversation optimizes it.
Consequence: You've validated an opinion rather than run a diagnosis, and the expensive assumption inside it goes unexamined.
Fix: Run Q2 even when the user arrives certain. It takes two minutes and it either confirms them or saves them a year.
Pitfall 2: Letting the Count Replace the Pattern
Symptom: "4 out of 7, so you're in decline."
Consequence: Four yeses clustered on channel and price mean something completely different from four on support cost and architecture. The count alone routes people wrong.
Fix: Always name which questions came back yes, and say what that cluster means.
Pitfall 3: Accepting the Internal Trigger at Face Value
Symptom: "Manufacturing wants the line" becomes "the product should be retired."
Consequence: You run a full retirement when an End of Sale, a price change, or a repackage would have satisfied the actual need more cheaply.
Fix: Name the trigger as internal, then ask what specifically it needs. Often the cheaper move delivers it.
Pitfall 4: Skipping the Extension Test on an Obvious Case
Symptom: The product is clearly dying, so Q4 gets waved through.
Consequence: No written record of why extension failed. In three months someone asks, and the answer is a shrug.
Fix: Ask all four regardless. On a genuinely dead product it takes ninety seconds and produces a defensible "no, because…"
Pitfall 5: Treating "Nothing Yet" as a Non-Answer
Symptom: Every run produces a play, because producing a play feels like producing value.
Consequence: Healthy mature products get projects they didn't need, funded from budget that had somewhere better to be.
Fix: 0-1 yeses means no play. Say so, set a review date, and stop.
References
Related Skills
These stand on their own — none is a prerequisite for this skill, and this skill isn't a prerequisite for them.
product-lifecycle-plays— the framework behind this triage: the strategy grid, the seven hazards, the risk register, the portfolio worksheeteol-readiness-advisor— the retirement play's go/no-goeol-process— running a retirement end to endorganic-growth-advisor— which growth path an extension servesansoff-matrix— where the next tranche of growth comes fromworkshop-facilitation— the interaction protocol
External Frameworks
- Product Life Cycle (PLC) — the five-stage economic model
- PLC strategy grid — marketing-mix levers by stage
Provenance
- Distilled from practitioner experience running product lifecycle transitions and product retirements across software, hardware, and regulated industries.
Dépôt GitHub
Questions fréquentes
Qu’est-ce que le Skill lifecycle-play-advisor ?
lifecycle-play-advisor est un Skill Claude créé par deanpeters. Un Skill regroupe des instructions et des ressources que Claude charge à la demande pour effectuer des tâches liées à lifecycle-play-advisor sans consigne supplémentaire.
Comment installer lifecycle-play-advisor ?
Utilisez les commandes d’installation de cette page : ajoutez lifecycle-play-advisor à Claude Code comme plugin ou clonez son dépôt dans votre dossier skills, puis redémarrez Claude pour charger le Skill.
À quelle catégorie appartient lifecycle-play-advisor ?
lifecycle-play-advisor appartient à la catégorie Autre.
lifecycle-play-advisor est-il gratuit ?
Oui. lifecycle-play-advisor est référencé sur AIMCP et son installation est gratuite.
Compétences associées
LlamaGuard est le modèle de Meta, doté de 7 à 8 milliards de paramètres, conçu pour modérer les entrées et sorties des LLM selon six catégories de sécurité comme la violence et les discours haineux. Il offre une précision de 94 à 95 % et peut être déployé avec vLLM, Hugging Face ou Amazon SageMaker. Utilisez cette compétence pour intégrer facilement le filtrage de contenu et des garde-fous de sécurité dans vos applications d'IA.
Cette compétence de Claude aide les développeurs à optimiser les coûts du cloud grâce au redimensionnement des ressources, aux stratégies d'étiquetage et à l'analyse des dépenses. Elle fournit un cadre pour réduire les dépenses cloud et mettre en œuvre une gouvernance des coûts sur AWS, Azure et GCP. Utilisez-la lorsque vous devez analyser les coûts d'infrastructure, redimensionner les ressources ou respecter des contraintes budgétaires.
Cette compétence Claude analyse les marchés des paris sportifs, incluant les spreads, les over/under et les paris spéciaux, en examinant les tendances historiques et les statistiques situationnelles pour identifier les paris à valeur ajoutée. Elle fournit une sortie en markdown structuré avec des recommandations actionnables à des fins éducatives. Les développeurs doivent l'utiliser pour des outils d'analyse de paris sportifs tout en notant qu'elle est conçue uniquement pour le divertissement et l'éducation.
Cette compétence quantifie les LLMs en précision 8 bits ou 4 bits à l'aide de bitsandbytes, permettant une réduction de 50 à 75 % de la mémoire utilisée avec une perte de précision minime. Elle est idéale pour exécuter des modèles plus volumineux sur une mémoire GPU limitée ou pour accélérer l'inférence, prenant en charge des formats comme INT8, NF4 et FP4. La compétence s'intègre à HuggingFace Transformers et permet l'entraînement QLoRA ainsi que l'utilisation d'optimiseurs en 8 bits.
