Conciseness

The body is dense with concrete domain data Claude does not know (discount ranges, retry schedules, recovery benchmarks, offer-to-reason mappings), so most tokens earn their place. It is not a 5 because of trimmable material: the ~40-line ASCII UI mockup, the 'You are an expert in...' role preamble, and some table rows that restate the referenced files' content.

Actionability

For an instruction-only skill the guidance is fully executable: exact parameters ('20-30% off for 2-3 months,' 'Retry 1: 24 hours after failure... Retry 4: 7 days'), a day-by-day dunning email sequence with tone and content, a complete offer-to-reason mapping table, and a worked health-score formula with a partial-coverage example. Per the rubric's scoring notes, absence of code is not penalized when the guidance is this actionable.

Workflow Clarity

Multi-step processes are clearly sequenced ('Trigger → Survey → Dynamic Offer → Confirmation → Post-Cancel' with five numbered steps; 'Pre-dunning → Smart retry → Dunning emails → Grace period → Hard cancel'), with some validation checkpoints ('Validate score mappings and intervention thresholds against later observed renewal/churn outcomes,' A/B testing with metrics). It is not a 5 because the three build/optimize/set-up modes lack an explicit post-launch verification step tying implementation back to the stated benchmarks (e.g., measure save rate against the 25-35% target after deploying).

Progressive Disclosure

The body is a well-sectioned overview with two real, clearly signaled, one-level-deep references ('For detailed cancel flow patterns by industry and billing provider, see references/cancel-flow-patterns.md'; 'For the complete dunning playbook with provider-specific setup, see references/dunning-playbook.md'), and industry/provider detail is correctly split out. It is not a 5 because some inlined content (the ASCII mockup, the full UI principles list, the tool comparison tables) reads as reference material that belongs in the bundle files.

Specificity

The description lists several specific, concrete actions — 'build cancellation flows,' 'set up save offers,' 'recover failed payments' — giving clear capability coverage. It falls short of the 5 anchor because 'reduce churn' and 'implement retention strategies' are generic goal statements rather than concrete capabilities, leaving minor gaps.

Completeness

It explicitly answers both 'what' (build cancellation flows, set up save offers, recover failed payments, implement retention strategies) and 'when' ('Also use when the user mentions...' followed by a long list of concrete trigger phrases, plus 'Use this whenever someone is losing subscribers'). This matches the 5 anchor exactly.

Trigger Term Quality

Comprehensive coverage of natural terms including technical synonyms ('churn,' 'cancel flow,' 'dunning,' 'win-back,' 'offboarding,' 'involuntary churn,' 'exit survey') and real user phrasings ('people keep canceling,' 'churn rate is too high,' 'how do I keep users,' 'customers are leaving'). This matches the 5 anchor's requirement for synonyms and natural variations.

Distinctiveness Conflict Risk

The SaaS retention/churn niche is distinct with domain-specific triggers, and it explicitly delegates boundary cases to sibling skills ('For post-cancel win-back email sequences, see emails. For in-app upgrade paywalls, see paywalls.'), minimizing conflict risk. Not 4: it goes beyond merely being mostly distinct by actively disambiguating the closest overlapping skills.