内容选题与增长策略

问题

内容团队常见的卡点不是不会写，而是选题来源太散：关键词表、销售录音、工单、论坛和竞品博客同时在叫，最后变成什么热就追什么。结果是栏目边界模糊，/blog 里既有泛流量文章，也有零散 case study，读者和搜索引擎都很难判断站点的专业主题。这个技能要把这些信号收敛成可用的内容策略：先明确业务目标与客户语言，再判断一篇内容更偏向 searchable 还是 shareable，然后落到栏目、主题簇和优先选题上。

工作方式

使用时，技能会先检查是否存在 .agents/product-marketing.md（或旧配置中的 product-marketing-context.md）。如果已有产品市场背景，它会先读取，再只追问缺失或与当前任务强相关的信息；没有则按业务、客户研究、现有内容与资源、竞品格局四个维度补齐上下文，例如目标客户、购买前问题、异议、支持工单主题、现有梯队和竞品内容空档。

核心步骤包括：

• 区分流量来源：searchable 抓已有搜索需求，优先保证关键词、搜索意图、标题和结构清晰；shareable 负责创造观点、数据和传播动机。
• 建立栏目框架：从产品问题、受众学习需求、搜索量和竞品排名中提炼 3-5 个 content pillars，再为每个 pillar 拆出主题簇；大多数内容留在 /blog 并通过内链连接，只有深度系统的主题才考虑独立 hub 或 guide。
• 按购买阶段找题：用 what is、best、vs、pricing、templates、how to use 等修饰词映射 awareness / consideration / decision / implementation 阶段。
• 从研究信号收题：如果提供 Ahrefs、SEMrush 或 GSC 导出，会按 topic cluster、buyer stage、意图、quick wins 和 content gaps 整理；如果提供销售或客服记录，会抽取问题、异议、原话和竞品对比；也可以结合 Reddit、Quora、Indie Hackers、HN、Product Hunt 等论坛搜索补足真实语言。
• 给创意打分：按 Customer Impact 40%、Content-Market Fit 30%、Search Potential 20%、Resource Requirements 10% 输出可排序的选题表，并说明推荐类型是 use-case、hub/spoke、thought leadership、data-driven、roundup 还是 case study。

适用边界

它适合规划层：当你需要决定写什么、放在哪个栏目、服务哪个搜索阶段或为什么值得传播时，产出最有价值。它不会替代 seo-audit 的技术性 on-page 检查，也不会直接执行 copywriting 的成稿工作；对 programmatic-seo 的规模化内容生产，只负责策略侧衔接。若缺少产品定位、客户访谈或关键词数据，它仍可基于已有信息产出草案，但口味会明显依赖输入质量；搜索量、难度和优先级应结合你的真实数据校准，不要把它当成排名保证。

How it works

The skill starts by checking for existing product-marketing context at .agents/product-marketing.md, .claude/product-marketing.md, or the legacy product-marketing-context.md. It reads that first and only asks for what is missing or task-specific. Without existing context, it gathers positioning, ICP, content goal, customer questions, objections, support themes, current resources, formats, and competitor coverage.

It then turns that research into strategy:

• Separate demand capture from demand creation: searchable content targets an exact keyword or question, clear intent, strong titles, headings, and comprehensive coverage; shareable content leads with a novel insight, original data, counterintuitive take, useful story, or honest behind-the-scenes lesson.
• Build pillars and clusters: identify 3-5 pillars through product problems, audience needs, search volume, and competitor activity; fan each pillar into related subtopics.
• Map keywords to the buyer journey: use awareness, consideration, decision, and implementation modifiers so topics reach the right intent, from what is and guide to through best, vs, pricing, templates, and how to use.
• Mine idea sources: keyword exports become prioritized tables by cluster, buyer stage, intent, quick wins, and content gaps; call transcripts and survey data surface pain points, objections, voice-of-customer phrases, and resource requests; forum searches like site:reddit.com and site:quora.com surface recurring questions; competitor URLs reveal covered topics, top formats, and weak spots.
• Score and prioritize: it weights Customer Impact at 40%, Content-Market Fit at 30%, Search Potential at 20%, and Resource Requirements at 10%, then returns content pillars, priority topics with keyword/stage/type, customer-research backing, and a topic-cluster map.

When to use it

Use it at the planning layer: deciding what to publish, how topics group into columns, which stage a topic serves, or whether an idea should chase search or earn sharing. It is not a replacement for technical seo-audit, general article writing, or scaled generation through programmatic-seo; use it before those workflows so the execution work has a clear editorial map.

Output quality is bounded by input quality. Product knowledge, customer language, sales patterns, ticket themes, and keyword data should be as real and recent as possible. If no exports are provided, it can still structure a draft from available input, but priority scores, search potential, and resource estimates must be validated against your own analytics and capacity before committing an editorial calendar.