[{"data":1,"prerenderedAt":353},["ShallowReactive",2],{"blog-keyword-library-maintenance":3,"blog-prev-keyword-library-maintenance":122,"blog-next-keyword-library-maintenance":238},{"id":4,"title":5,"author":6,"body":7,"canonical":106,"category":107,"cover":106,"date":108,"description":109,"draft":110,"extension":111,"keywords":106,"meta":112,"navigation":113,"noindex":110,"ogImage":106,"path":114,"seo":115,"seoDescription":106,"seoTitle":106,"stem":116,"tags":117,"updated":106,"__hash__":121},"blog\u002Fblog\u002Fkeyword-library-maintenance.md","词库半年不维护就废了，一半的词已经没人搜","元析智能",{"type":8,"value":9,"toc":93},"minimark",[10,14,18,21,24,27,30,33,36,39,42,45,48,51,54,57,60,63,66,69,72,75,78,81,84,87,90],[11,12,13],"p",{},"词库半年不维护就废了。2026 年 4 月我们把 2025 年做的词表翻出来重看，870 个词里，210 个已经写过，130 个连续 6 个月零曝光，60 个词的意思已经变了。",[15,16,17],"h2",{"id":17},"半年里会发生什么",[11,19,20],{},"三件事。新词出现，产品迭代、行业换说法都会带来新词。旧词消失，功能下线、用户改叫法。词义漂移，同一个词两年前是学习意图，现在是采购意图。",[11,22,23],{},"不维护的词库，一半是过期信息，用起来比没有更危险。",[15,25,26],{"id":26},"每月做什么",[11,28,29],{},"30 分钟，三件事：加新词、标已写的、删连续两月零曝光的。",[11,31,32],{},"新词来源固定三个：搜索后台的新查询、工单新问题、销售提的新叫法。",[15,34,35],{"id":35},"每季度做什么",[11,37,38],{},"重算一次优先级。竞争格局会变，原来好打的词可能被别人做掉了，原来难的可能空出来。",[11,40,41],{},"我们那次季度重算，120 个词里 31 个分数变化超过 20 分。",[15,43,44],{"id":44},"每年做什么",[11,46,47],{},"重建一次。把全年写过的内容倒推回去，看哪些词表里的词真的被写成了页面，哪些没写。",[11,49,50],{},"2026 年 4 月那次倒推发现，870 个词里只有 210 个变成了内容，占 24%。剩下 76% 是白挖的，说明挖词阶段的筛选太松。",[15,52,53],{"id":53},"谁来维护",[11,55,56],{},"写内容的人自己维护，别外包。词库是判断工具，不是资料库，交给别人维护，判断就断了。",[11,58,59],{},"我们现在的做法是每周五下班前 30 分钟，加每月第一个周一 30 分钟重算。",[15,61,62],{"id":62},"一个更狠的做法",[11,64,65],{},"每季度直接把词库里三个月没动过的词删掉。删了不心疼，说明它们本来也不重要。",[11,67,68],{},"我们试过一轮，删掉 300 多个词，后来真正被找回来的不到 20 个。",[15,70,71],{"id":71},"最后",[15,73,74],{"id":74},"一个维护中的坑",[11,76,77],{},"别让两个人都能改同一张表而不留记录。我们有过一次，两个人同时整理，把对方的标签覆盖了，一周的工作白做。",[11,79,80],{},"后来改成每周只有一个人有编辑权，其他人提交申请。",[15,82,83],{"id":83},"怎么判断该重算了",[11,85,86],{},"看三个信号：上季度的选题命中率、新词出现速度、有没有词从第 30 名外突然进前 10。",[11,88,89],{},"出现两个，就该重算，不用等季度末。",[11,91,92],{},"词库的价值不在大，在准。一个 200 词的库，每个词都有承接页，比 2000 个词的库有用得多。",{"title":94,"searchDepth":95,"depth":95,"links":96},"",2,[97,98,99,100,101,102,103,104,105],{"id":17,"depth":95,"text":17},{"id":26,"depth":95,"text":26},{"id":35,"depth":95,"text":35},{"id":44,"depth":95,"text":44},{"id":53,"depth":95,"text":53},{"id":62,"depth":95,"text":62},{"id":71,"depth":95,"text":71},{"id":74,"depth":95,"text":74},{"id":83,"depth":95,"text":83},null,"SEO","2026-11-06","2026 年 4 月我们重看 2025 年的词表，870 个词里 210 个已写、130 个零曝光、60 个词义变了。词库不是建一次就完事，我们固定每月小更新、每季度重算。",false,"md",{},true,"\u002Fblog\u002Fkeyword-library-maintenance",{"title":5,"description":109},"blog\u002Fkeyword-library-maintenance",[118,119,120],"词库搭建","关键词研究","内容运营","G72W5tGDPZLVAHvuvMgoHBn-HNsd0T-zbQSXWK-RCtg",{"id":123,"title":124,"author":6,"body":125,"canonical":106,"category":120,"cover":106,"date":227,"description":228,"draft":110,"extension":111,"keywords":106,"meta":229,"navigation":113,"noindex":110,"ogImage":106,"path":230,"seo":231,"seoDescription":106,"seoTitle":106,"stem":232,"tags":233,"updated":106,"__hash__":237},"blog\u002Fblog\u002Fcontent-half-life.md","内容的半衰期：哪些文章三年后还在带量",{"type":8,"value":126,"toc":217},[127,130,133,136,139,142,145,148,151,154,157,160,163,166,169,172,175,178,181,184,187,190,193,196,199,202,205,208,211,214],[11,128,129],{},"内容有半衰期，有的三个月就没流量，有的三年还在涨。我们跟踪了 120 篇内容的流量曲线，分出三类：快消型、稳增型、长尾型。",[15,131,132],{"id":132},"什么是内容半衰期",[11,134,135],{},"一篇文章的月访问量掉到峰值的 50%，所花的时间，就是它的半衰期。",[11,137,138],{},"半衰期短的内容，需要持续补充；半衰期长的内容，是资产。",[15,140,141],{"id":141},"三类半衰期",[11,143,144],{},"快消型：半衰期 1 到 3 个月。新闻、热点分析、活动内容。写完就得推，过了时间就没用。",[11,146,147],{},"稳增型：半衰期 12 到 18 个月。教程、选型、对比。前期爬坡慢，稳定后有持续流量。",[11,149,150],{},"长尾型：半衰期 3 年以上。概念解释、踩坑记录、方法论。流量不大，但一直有。",[15,152,153],{"id":153},"我们跟踪的数据",[11,155,156],{},"2024 年 9 月开始，我们跟踪 120 篇文章 2 年：快消型 34 篇，平均半衰期 1.8 个月；稳增型 61 篇，平均 14 个月；长尾型 25 篇，两年后流量是峰值的 70%。",[11,158,159],{},"有意思的是，长尾型内容单篇流量最低，但两年累计流量最高，因为它在持续带量。",[15,161,162],{"id":162},"什么内容半衰期长",[11,164,165],{},"第一，回答\"不会变的问题\"。什么是 API、为什么需要限流，这些十年后还有人问。",[11,167,168],{},"第二，有独家信息的内容。别人抄不走，所以能长期占位。",[11,170,171],{},"第三，结构完整、可被引用的内容。AI 搜索和同行引用会持续带来流量。",[15,173,174],{"id":174},"怎么用半衰期做排期",[11,176,177],{},"快消型：控制在 20% 以内。做多了，团队会很累，而且不积累。",[11,179,180],{},"稳增型：主力，占 50%。这是搜索流量的基本盘。",[11,182,183],{},"长尾型：占 30%。短期看不到效果，但每年都在积累。",[11,185,186],{},"我们的季度排期一般按 2:5:3 分，和流量贡献的节奏匹配。",[15,188,189],{"id":189},"一个提醒",[11,191,192],{},"半衰期会变长。一篇教程如果每年更新，半衰期可以从 14 个月拉到 3 年。",[11,194,195],{},"所以\"更新\"不只是维护，还是延长资产寿命。",[15,197,198],{"id":198},"两个陷阱",[11,200,201],{},"第一个陷阱，把快消内容当资产。热点文章写完就完了，别指望它持续带量，也别为它做大量内链。",[11,203,204],{},"第二个陷阱，用快消数据判断团队。如果团队这个月写了两篇热点，流量好看，但三个月后归零。考核要看长尾型内容的存量。",[11,206,207],{},"我们的做法是分两个指标：当月流量看新内容，年度流量看资产内容。",[15,209,210],{"id":210},"一句判断",[11,212,213],{},"做内容不能只看当月流量。快消型内容让你这个月好看，长尾型内容让你三年后还有饭吃。",[11,215,216],{},"两者都要有，但比例别搞反。我们的比例是 2:5:3，跑了两年，比 5:3:2 稳。这个比例每年调一次，看当年的数据再定。",{"title":94,"searchDepth":95,"depth":95,"links":218},[219,220,221,222,223,224,225,226],{"id":132,"depth":95,"text":132},{"id":141,"depth":95,"text":141},{"id":153,"depth":95,"text":153},{"id":162,"depth":95,"text":162},{"id":174,"depth":95,"text":174},{"id":189,"depth":95,"text":189},{"id":198,"depth":95,"text":198},{"id":210,"depth":95,"text":210},"2026-11-05","内容有半衰期，有的三个月就没流量，有的三年还在涨。我们跟踪了 120 篇内容的流量曲线，分出三类：快消型、稳增型、长尾型。半衰期怎么判断、怎么用它做排期都在这篇，数据来自我们对 120 篇内容的两年跟踪。",{},"\u002Fblog\u002Fcontent-half-life",{"title":124,"description":228},"blog\u002Fcontent-half-life",[234,235,236],"内容评估","内容策略","数据","q-rbxJmVri6Tw5XjEVR2NSeOO82TWzmDEap-WFPvTak",{"id":239,"title":240,"author":6,"body":241,"canonical":106,"category":107,"cover":106,"date":342,"description":343,"draft":110,"extension":111,"keywords":106,"meta":344,"navigation":113,"noindex":110,"ogImage":106,"path":345,"seo":346,"seoDescription":106,"seoTitle":106,"stem":347,"tags":348,"updated":106,"__hash__":352},"blog\u002Fblog\u002Fcitation-click-landing.md","被 AI 引用之后，用户点进来落在哪一页",{"type":8,"value":242,"toc":333},[243,246,249,252,255,258,261,264,267,270,274,277,280,283,287,290,293,296,300,303,306,309,312,315,318,321,324,327,330],[11,244,245],{},"2026 年 11 月我们查一个客户的 AI 引用数据，发现有内容的页面被引用最多，但那些页面的留资几乎为零。引用带来了访问，访问走到了死路。",[15,247,248],{"id":248},"引用点击会落在哪",[11,250,251],{},"大部分情况落在被引用的那个页面，因为用户点的是答案里的来源链接。",[11,253,254],{},"少数情况用户不会点，而是看完回答直接搜品牌名。这部分流量会被记成品牌词搜索或直接访问。",[11,256,257],{},"两种情况都要求被引用页面本身能承接。",[15,259,260],{"id":260},"承接断在哪",[11,262,263],{},"被引用的页面通常是文档、博客、术语解释，这类页面的设计目标是讲清楚一个问题，不是促成转化。",[11,265,266],{},"于是出现一个断点：用户读完，觉得有用，但不知道下一步做什么。",[11,268,269],{},"页面上没有入口、没有相关产品、没有更多资料，用户就走了。",[15,271,273],{"id":272},"第一步找到被引用的页面","第一步：找到被引用的页面",[11,275,276],{},"从引用记录里整理页面清单，按被引用次数排序。",[11,278,279],{},"再和流量数据交叉，看这些页面实际有多少访问。",[11,281,282],{},"两个数据都靠前的页面，优先改。",[15,284,286],{"id":285},"第二步加一个不打扰的下一步","第二步：加一个不打扰的下一步",[11,288,289],{},"在文章末尾加一句，指向更详细的内容或者试用入口，比如\"想看完整的配置步骤，这里有文档\"。",[11,291,292],{},"在正文中间插一个轻量入口，比如\"这一节涉及的具体参数，我们的产品页有完整列表\"。",[11,294,295],{},"不要弹窗，不要遮内容。被引用页面的读者是来找答案的，打扰就流失。",[15,297,299],{"id":298},"第三步别把文档页改成销售页","第三步：别把文档页改成销售页",[11,301,302],{},"这是最容易犯的错。看到文档页流量大，就往上加各种 CTA 和表单。",[11,304,305],{},"结果内容变得难读，引用减少，流量也没了。",[11,307,308],{},"我们的原则是内容主体不动，只在合适的节点加导航，加完之后自己读一遍，不别扭才算过。",[15,310,311],{"id":311},"怎么测量",[11,313,314],{},"AI 引用的来源流量很难精确归因，因为很多引擎不传 referrer。",[11,316,317],{},"我们的做法是间接看三个数：被引用页面的总访问、这些页面的站内跳转率、品牌词搜索变化。",[11,319,320],{},"站内跳转率是最直接的指标。加了下一步之后，从文档页跳到产品页的比例应该上升。",[15,322,323],{"id":323},"一个态度",[11,325,326],{},"GEO 的终点不是被引用，是被引用之后有人愿意继续了解你。",[11,328,329],{},"引用是入口，承接是转化。只做引用不做承接，等于把流量引到一个没有门的房间。",[11,331,332],{},"这件事的投入很小，改几个页面的末尾就行。但它决定了 GEO 的投入能不能变成生意。",{"title":94,"searchDepth":95,"depth":95,"links":334},[335,336,337,338,339,340,341],{"id":248,"depth":95,"text":248},{"id":260,"depth":95,"text":260},{"id":272,"depth":95,"text":273},{"id":285,"depth":95,"text":286},{"id":298,"depth":95,"text":299},{"id":311,"depth":95,"text":311},{"id":323,"depth":95,"text":323},"2026-11-07","引用带来的点击通常落在被引用页面，而那一页往往没有转化入口。这篇写承接断在哪、怎么在不打扰阅读的前提下加下一步，以及怎么测量。",{},"\u002Fblog\u002Fcitation-click-landing",{"title":240,"description":343},"blog\u002Fcitation-click-landing",[349,350,351],"GEO","转化路径","落地页","59VysvmXMIaHC1faOdEZ615TAhHXhzDH2Ig14Ey9ZV4",1794088099733]