[{"data":1,"prerenderedAt":347},["ShallowReactive",2],{"blog-content-half-life":3,"blog-prev-content-half-life":130,"blog-next-content-half-life":239},{"id":4,"title":5,"author":6,"body":7,"canonical":114,"category":115,"cover":114,"date":116,"description":117,"draft":118,"extension":119,"keywords":114,"meta":120,"navigation":121,"noindex":118,"ogImage":114,"path":122,"seo":123,"seoDescription":114,"seoTitle":114,"stem":124,"tags":125,"updated":114,"__hash__":129},"blog\u002Fblog\u002Fcontent-half-life.md","内容的半衰期：哪些文章三年后还在带量","元析智能",{"type":8,"value":9,"toc":102},"minimark",[10,14,18,21,24,27,30,33,36,39,42,45,48,51,54,57,60,63,66,69,72,75,78,81,84,87,90,93,96,99],[11,12,13],"p",{},"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。",[15,16,17],"h2",{"id":17},"什么是内容半衰期",[11,19,20],{},"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。",[11,22,23],{},"半衰期短的内容，需要持续补充；半衰期长的内容，是资产。",[15,25,26],{"id":26},"三类半衰期",[11,28,29],{},"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。",[11,31,32],{},"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。",[11,34,35],{},"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。",[15,37,38],{"id":38},"我们跟踪的数据",[11,40,41],{},"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%。",[11,43,44],{},"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。",[15,46,47],{"id":47},"什么内容半衰期长",[11,49,50],{},"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。",[11,52,53],{},"第二，有独家信息的内容。别人抄不走，所以能长期占位。",[11,55,56],{},"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。",[15,58,59],{"id":59},"怎么用半衰期做排期",[11,61,62],{},"快消型：控制在 20% 以内。做多了，团队会很累，而且不积累。",[11,64,65],{},"稳增型：主力，占 50%。这是搜索流量的基本盘。",[11,67,68],{},"长尾型：占 30%。短期看不到效果，但每年都在积累。",[11,70,71],{},"我们的季度排期一般按 2:5:3 分，和流量贡献的节奏匹配。",[15,73,74],{"id":74},"一个提醒",[11,76,77],{},"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。",[11,79,80],{},"所以\"更新\"不只是维护，还是延长资产寿命。",[15,82,83],{"id":83},"两个陷阱",[11,85,86],{},"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。",[11,88,89],{},"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。",[11,91,92],{},"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。",[15,94,95],{"id":95},"一句判断",[11,97,98],{},"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。",[11,100,101],{},"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":103,"searchDepth":104,"depth":104,"links":105},"",2,[106,107,108,109,110,111,112,113],{"id":17,"depth":104,"text":17},{"id":26,"depth":104,"text":26},{"id":38,"depth":104,"text":38},{"id":47,"depth":104,"text":47},{"id":59,"depth":104,"text":59},{"id":74,"depth":104,"text":74},{"id":83,"depth":104,"text":83},{"id":95,"depth":104,"text":95},null,"内容运营","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。",false,"md",{},true,"\u002Fblog\u002Fcontent-half-life",{"title":5,"description":117},"blog\u002Fcontent-half-life",[126,127,128],"内容评估","内容策略","数据","q-rbxJmVri6Tw5XjEVR2NSeOO82TWzmDEap-WFPvTak",{"id":131,"title":132,"author":6,"body":133,"canonical":114,"category":227,"cover":114,"date":228,"description":229,"draft":118,"extension":119,"keywords":114,"meta":230,"navigation":121,"noindex":118,"ogImage":114,"path":231,"seo":232,"seoDescription":114,"seoTitle":114,"stem":233,"tags":234,"updated":114,"__hash__":238},"blog\u002Fblog\u002Fchina-mobile-network-field-test.md","国内三大运营商 4G 实测：同一页能差几秒",{"type":8,"value":134,"toc":219},[135,138,141,144,147,150,153,156,159,162,165,168,171,174,177,180,183,186,189,192,195,198,201,204,207,210,213,216],[11,136,137],{},"2026 年 10 月我们做了一组国内移动网络实测，三个城市、三家运营商、同一台中端安卓机，测五个页面的首屏。同一页面最好和最差差了 1.7 秒。这个差异在报表里看不到，但用户在承受。",[15,139,140],{"id":140},"怎么测的",[11,142,143],{},"设备用同一台中端机，清空缓存，每次测之前重启浏览器。",[11,145,146],{},"每个城市选三个时段：早高峰、午间、晚上八点。每个组合测三次，取中位数。",[11,148,149],{},"页面选五个：首页、产品页、定价页、文档页、博客文章。覆盖不同类型。",[11,151,152],{},"测试方式是手机上开热点给笔记本，用浏览器开发者工具记录。这样比纯手机录屏更容易拿到细粒度数据。",[15,154,155],{"id":155},"测出来的差异",[11,157,158],{},"首字节时间，不同城市之间差 200 到 400 毫秒，同一城市不同运营商能差 300 毫秒。",[11,160,161],{},"完整加载时间差异更大。资源多的页面，在信号弱的地方能差到 1.7 秒。",[11,163,164],{},"晚上八点是普遍最差的时段，比午间慢三成左右。这个规律在三个城市都成立。",[15,166,167],{"id":167},"差异从哪来",[11,169,170],{},"第一，基站负载。晚高峰人多，每个用户的带宽被摊薄。",[11,172,173],{},"第二，跨网访问。某些运营商的用户访问其他运营商机房的资源，要多绕一跳。",[11,175,176],{},"第三，DNS 解析。不同运营商的 DNS 解析速度和结果不一样，有时会解析到较远的节点。",[11,178,179],{},"第四，CDN 节点覆盖。如果 CDN 在某家运营商没有节点，用户就要跨网取资源。",[15,181,182],{"id":182},"对官网的启示",[11,184,185],{},"第一，别只用 Wi-Fi 测。大部分验收测试都在办公室 Wi-Fi 下做，数据好看，不代表用户体感。",[11,187,188],{},"第二，移动端单独设验收线。我们的标准是 4G 环境下首屏 3 秒内，文档页 2.5 秒内。",[11,190,191],{},"第三，CDN 要看运营商覆盖。三家都有节点的服务商，跨网差异会小很多。",[11,193,194],{},"第四，晚高峰要单独测。日常测出来的数据，晚八点可能差三成。",[15,196,197],{"id":197},"怎么自己复现",[11,199,200],{},"不需要专门的设备。一台中端安卓机、一张不同运营商的卡、一个固定的测试页面就够了。",[11,202,203],{},"测的时候关闭其他应用，用浏览器的无痕模式，每次清缓存。",[11,205,206],{},"测完记录城市、运营商、时段、首屏时间四项，凑够二十组数据就能看出规律。",[15,208,209],{"id":209},"一个态度",[11,211,212],{},"性能优化的目标不是实验室里的满分，是真实用户在真实网络下的体验。",[11,214,215],{},"国内移动网络的复杂性远超大部分人的想象，运营商、地域、时段、机型四个变量交叉，差异可以很大。",[11,217,218],{},"官网的流量里移动端占一半以上，这条线不测，等于放弃了大部分用户的体感。",{"title":103,"searchDepth":104,"depth":104,"links":220},[221,222,223,224,225,226],{"id":140,"depth":104,"text":140},{"id":155,"depth":104,"text":155},{"id":167,"depth":104,"text":167},{"id":182,"depth":104,"text":182},{"id":197,"depth":104,"text":197},{"id":209,"depth":104,"text":209},"工具评测","2026-11-04","三个城市、三家运营商、同一台中端机，测同一批页面的首屏。差异最大的一组差了 1.7 秒。这篇写测试方法、差异来源，以及官网该按哪条线验收。",{},"\u002Fblog\u002Fchina-mobile-network-field-test",{"title":132,"description":229},"blog\u002Fchina-mobile-network-field-test",[235,236,237],"移动端","国内访问","实测","t6gL_SdgovqjuXqQkQUJ5U79LqlAk6akKvd8_rjHzD0",{"id":240,"title":241,"author":6,"body":242,"canonical":114,"category":336,"cover":114,"date":337,"description":338,"draft":118,"extension":119,"keywords":114,"meta":339,"navigation":121,"noindex":118,"ogImage":114,"path":340,"seo":341,"seoDescription":114,"seoTitle":114,"stem":342,"tags":343,"updated":114,"__hash__":346},"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",{"type":8,"value":243,"toc":325},[244,247,250,253,256,259,262,265,268,271,274,277,280,283,286,289,292,295,298,301,304,307,310,313,316,319,322],[11,245,246],{},"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。",[15,248,249],{"id":249},"半年里会发生什么",[11,251,252],{},"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。",[11,254,255],{},"不维护的词库，一半是过期信息，用起来比没有更危险。",[15,257,258],{"id":258},"每月做什么",[11,260,261],{},"30 分钟，三件事：加新词、标已写的、删连续两月零曝光的。",[11,263,264],{},"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。",[15,266,267],{"id":267},"每季度做什么",[11,269,270],{},"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。",[11,272,273],{},"我们那次季度重算，120 个词里 31 个分数变化超过 20 分。",[15,275,276],{"id":276},"每年做什么",[11,278,279],{},"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。",[11,281,282],{},"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。",[15,284,285],{"id":285},"谁来维护",[11,287,288],{},"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。",[11,290,291],{},"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。",[15,293,294],{"id":294},"一个更狠的做法",[11,296,297],{},"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。",[11,299,300],{},"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。",[15,302,303],{"id":303},"最后",[15,305,306],{"id":306},"一个维护中的坑",[11,308,309],{},"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。",[11,311,312],{},"后来改成每周只有一个人有编辑权，其他人提交申请。",[15,314,315],{"id":315},"怎么判断该重算了",[11,317,318],{},"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。",[11,320,321],{},"出现两个，就该重算，不用等季度末。",[11,323,324],{},"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":103,"searchDepth":104,"depth":104,"links":326},[327,328,329,330,331,332,333,334,335],{"id":249,"depth":104,"text":249},{"id":258,"depth":104,"text":258},{"id":267,"depth":104,"text":267},{"id":276,"depth":104,"text":276},{"id":285,"depth":104,"text":285},{"id":294,"depth":104,"text":294},{"id":303,"depth":104,"text":303},{"id":306,"depth":104,"text":306},{"id":315,"depth":104,"text":315},"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。",{},"\u002Fblog\u002Fkeyword-library-maintenance",{"title":241,"description":338},"blog\u002Fkeyword-library-maintenance",[344,345,115],"词库搭建","关键词研究","G72W5tGDPZLVAHvuvMgoHBn-HNsd0T-zbQSXWK-RCtg",1794088099738]