[{"data":1,"prerenderedAt":336},["ShallowReactive",2],{"blog-data-post-structure":3,"blog-prev-data-post-structure":116,"blog-next-data-post-structure":223},{"id":4,"title":5,"author":6,"body":7,"canonical":100,"category":101,"cover":100,"date":102,"description":103,"draft":104,"extension":105,"keywords":100,"meta":106,"navigation":107,"noindex":104,"ogImage":100,"path":108,"seo":109,"seoDescription":100,"seoTitle":100,"stem":110,"tags":111,"updated":100,"__hash__":115},"blog\u002Fblog\u002Fdata-post-structure.md","数据文的结构：先给反常识结论，再交代口径","元析智能",{"type":8,"value":9,"toc":88},"minimark",[10,14,18,21,24,27,30,33,36,39,42,45,48,51,54,57,60,63,66,69,73,76,79,82,85],[11,12,13],"p",{},"数据文最容易写成报告：先讲背景，再列数据，最后给一堆结论。读者一般读三段就关掉。我们的结构是反过来的：第一段就给那个最反常识的结论，然后才解释数据从哪来。",[15,16,17],"h2",{"id":17},"第一段只做一件事",[11,19,20],{},"给一个反常识的结论，并且带上数字。比如“我们把 1260 个长尾词按搜索量排序，前 30 个选题里只有 4 个带来了试用”。",[11,22,23],{},"这一段决定读者要不要继续。数据文的价值在前 100 字，不在后面的图表。",[15,25,26],{"id":26},"第二段交代来源和口径",[11,28,29],{},"样本多少、时间范围、怎么统计、有哪些已知偏差。这一段会让文章显得可信，也是别人愿意引用的原因。",[11,31,32],{},"口径要写在正文里，不要塞进脚注。AI 摘录时看不到脚注。",[15,34,35],{"id":35},"第三段给两三个细分发现",[11,37,38],{},"不要给十个。挑三个最有用、最能支撑结论的。每个发现配一个具体例子。",[11,40,41],{},"我们的习惯是：一个发现对应一句“所以该怎么做”。只有数据没有行动建议，读者读完会觉得看了一场热闹。",[15,43,44],{"id":44},"第四段写适用边界",[11,46,47],{},"什么情况下这个结论不成立。这一段最容易被省，也最加分。",[11,49,50],{},"写清边界，说明你真的理解数据，不是拿几个数字凑观点。",[15,52,53],{"id":53},"图表放几张",[11,55,56],{},"两三张就够。每张图只讲一件事，标题写结论，不写“图 1 数据分布”这种。",[11,58,59],{},"图里的数字要能被正文引用，别放一张读者看不懂的复杂图。",[15,61,62],{"id":62},"一个具体的对比",[11,64,65],{},"同一批数据，我们写过两版。第一版按“数据概览、分布、趋势、结论”排，阅读完成率 34%。第二版按“结论、口径、发现、边界”排，完成率 61%。",[11,67,68],{},"内容没变，顺序变了。",[15,70,72],{"id":71},"一个时间上的提醒","## 一个时间上的提醒",[11,74,75],{},"数据文有保鲜期。我们 2026 年 3 月发的那篇用了 2025 年全年数据，到 10 月就该更新了。数据文的日期比观点文更敏感，读者会看数据到哪一天。",[15,77,78],{"id":78},"数据文的两个风险",[11,80,81],{},"一是样本太小却下大结论。我们的规矩是样本少于 30 个不出百分比结论。",[11,83,84],{},"二是数据过时。数据文里要写清数据截止时间，超过一年就该更新或者下架。",[11,86,87],{},"写清数据截止时间这一行，也能避免读者把旧结论当新结论用。",{"title":89,"searchDepth":90,"depth":90,"links":91},"",2,[92,93,94,95,96,97,98,99],{"id":17,"depth":90,"text":17},{"id":26,"depth":90,"text":26},{"id":35,"depth":90,"text":35},{"id":44,"depth":90,"text":44},{"id":53,"depth":90,"text":53},{"id":62,"depth":90,"text":62},{"id":71,"depth":90,"text":72},{"id":78,"depth":90,"text":78},null,"内容运营","2026-08-22","数据文不是把数字堆出来。我们固定四段结构：反常识结论、数据来源和口径、两三个细分发现、适用边界。顺序错了，读者读三段就关掉。",false,"md",{},true,"\u002Fblog\u002Fdata-post-structure",{"title":5,"description":103},"blog\u002Fdata-post-structure",[112,113,114],"数据文","内容结构","信息增益","nXnHM-VLBbFYaF9iFo6xNcVYQtEcFQDCImOG6MNlBWk",{"id":117,"title":118,"author":6,"body":119,"canonical":100,"category":211,"cover":100,"date":212,"description":213,"draft":104,"extension":105,"keywords":100,"meta":214,"navigation":107,"noindex":104,"ogImage":100,"path":215,"seo":216,"seoDescription":100,"seoTitle":100,"stem":217,"tags":218,"updated":100,"__hash__":222},"blog\u002Fblog\u002Fmobile-desktop-cwv-gap.md","移动端比桌面慢三秒，差距不只在网络",{"type":8,"value":120,"toc":203},[121,124,127,130,133,136,139,142,145,148,151,155,158,161,164,167,170,173,176,179,182,185,188,191,194,197,200],[11,122,123],{},"2026 年 9 月我们测一个客户站，同一页面桌面 1.8 秒、移动端 5.1 秒。客户第一反应是移动网络慢。查下来，网络只占了一半，另一半是 CPU 和渲染。",[15,125,126],{"id":126},"四个差距来源",[11,128,129],{},"第一，CPU。中端安卓机的性能大约是桌面电脑的四分之一，同样的 JS 在手机上执行时间翻几倍。",[11,131,132],{},"第二，网络。4G 的往返延迟比宽带高，带宽也不稳定，晚高峰更明显。",[11,134,135],{},"第三，渲染。移动端屏幕小，但 DOM 没变，布局计算和绘制的工作量并没有减少。",[11,137,138],{},"第四，资源。移动端视口窄，可能触发不同的图片尺寸和懒加载逻辑，加载的资源反而不一样。",[15,140,141],{"id":141},"网络不是主因",[11,143,144],{},"很多人把移动端慢归咎于网络，实测下来 4G 的带宽对大部分页面够用，真正拖时间的是延迟。",[11,146,147],{},"一个请求来回 100 毫秒，首屏要串行发五个请求，光是等待就是半秒。桌面宽带这个数字是 20 毫秒。",[11,149,150],{},"所以减少请求数量比压缩资源体积更有效，尤其是在移动端。",[15,152,154],{"id":153},"cpu-降速才是大头","CPU 降速才是大头",[11,156,157],{},"桌面电脑跑得欢的 JS，在中端手机上执行时间能翻四倍。",[11,159,160],{},"如果首屏要执行 800 毫秒的 JS，手机上就是三秒多。这还没算解码图片和渲染布局。",[11,162,163],{},"测试时必须打开 CPU 降速。Lighthouse 默认给移动端 4 倍降速，这个值接近真实的中端机，别关掉。",[15,165,166],{"id":166},"移动端专属的四个坑",[11,168,169],{},"第一个，懒加载触发太早。移动端视口窄，图片懒加载的阈值如果按桌面设置，用户滚动一下就要下载好几张。",[11,171,172],{},"第二个，字体太大。移动端带宽和 CPU 都紧张，几兆的中文字体在手机上是灾难。",[11,174,175],{},"第三个，浮层太多。客服、Cookie 提示、下载 App 的横幅，三个浮层叠在小屏上，遮挡内容还影响交互。",[11,177,178],{},"第四个，横向溢出。页面出现横向滚动条，用户在手机上左右晃，体验直接崩掉。",[15,180,181],{"id":181},"怎么测",[11,183,184],{},"先用 DevTools 的移动端模拟，开 4G 降速和 4 倍 CPU 降速，看首屏指标。",[11,186,187],{},"再用真机测。模拟器和真机差很多，尤其是低端安卓机上的表现。",[11,189,190],{},"最后看 CrUX 按设备分组的数据。移动端的真实用户分布才是判断用户感受的依据。",[15,192,193],{"id":193},"一个判断",[11,195,196],{},"移动端优化不能靠桌面数据推算。同一套代码，桌面合格、移动端超标是常态。",[11,198,199],{},"我们的项目验收标准里，移动端的 LCP 单独设一条线，不和桌面共用。",[11,201,202],{},"因为大部分软件公司的流量里，移动端占比已经超过一半，甚至更高。只优化桌面，等于只做了一半。",{"title":89,"searchDepth":90,"depth":90,"links":204},[205,206,207,208,209,210],{"id":126,"depth":90,"text":126},{"id":141,"depth":90,"text":141},{"id":153,"depth":90,"text":154},{"id":166,"depth":90,"text":166},{"id":181,"depth":90,"text":181},{"id":193,"depth":90,"text":193},"技术SEO","2026-08-21","同一页面桌面 1.8 秒、移动 5.1 秒。很多人以为差在网络，其实 CPU 降速和布局复杂度占大头。这篇写四个差距来源和移动端专属的测试方法。",{},"\u002Fblog\u002Fmobile-desktop-cwv-gap",{"title":118,"description":213},"blog\u002Fmobile-desktop-cwv-gap",[219,220,221],"移动端","CoreWebVitals","测速","015VaLRnQmI9rnAktRBsKlBKBNOLm6Y8AltoddAZV7Y",{"id":224,"title":225,"author":6,"body":226,"canonical":100,"category":101,"cover":100,"date":326,"description":327,"draft":104,"extension":105,"keywords":100,"meta":328,"navigation":107,"noindex":104,"ogImage":100,"path":329,"seo":330,"seoDescription":100,"seoTitle":100,"stem":331,"tags":332,"updated":100,"__hash__":335},"blog\u002Fblog\u002Flongtail-aggregation.md","长尾词聚合：一个页面覆盖 20 个词，怎么写才不是缝合怪",{"type":8,"value":227,"toc":317},[228,231,234,237,240,243,246,249,252,255,259,262,265,268,271,275,278,281,284,287,290,293,296,299,302,305,308,311,314],[11,229,230],{},"长尾词一个个写文章成本太高，聚合到一个页面是常见做法。但聚合不等于把词堆进去。我们判断\"该不该聚合\"看三个条件，写法上分四层。",[15,232,233],{"id":233},"什么是长尾词聚合",[11,235,236],{},"把 10 到 30 个语义相关、意图相近的长尾词，放在一个页面上覆盖，而不是一个词写一篇。",[11,238,239],{},"比如\"API 限流\"下面有：限流算法、限流阈值、限流报错、分布式限流、限流和熔断的区别。这些词单独写五篇，每篇都写不深；聚合成一篇长文，反而更完整。",[15,241,242],{"id":242},"什么时候该聚合",[11,244,245],{},"三个条件同时满足才聚合：",[11,247,248],{},"第一，词群意图相同。都是\"了解方案\"或者都是\"解决问题\"，不能混着买意图的词。",[11,250,251],{},"第二，单个词搜索量小。月搜 100 以下的长尾，单独成篇性价比低。",[11,253,254],{},"第三，词之间有逻辑关系。能顺着读者的理解路径排下来，不是硬凑。",[15,256,258],{"id":257},"我们的写法四层结构","我们的写法：四层结构",[11,260,261],{},"第一层，开头给总答案。用 100 字回答\"这一组词背后的核心问题是什么\"。",[11,263,264],{},"第二层，按词群分节。每个 H2 对应 3 到 5 个长尾词，节内自然覆盖这些词，不刻意重复。",[11,266,267],{},"第三层，用表格和清单承载密集信息。多个词的对比用表格，步骤用清单，比大段文字清楚。",[11,269,270],{},"第四层，结尾给判断和下一步。告诉读者\"什么情况选哪个\"，把决策收住。",[15,272,274],{"id":273},"边界聚合页不能是缝合怪","边界：聚合页不能是缝合怪",[11,276,277],{},"三个红线：",[11,279,280],{},"第一，别把意图不同的词硬放一起。一个页面既讲\"是什么\"又讲\"怎么买\"，两边都做不好。",[11,282,283],{},"第二，别为了覆盖而覆盖。某个词和主题关系弱，就不收，宁可少覆盖两个。",[11,285,286],{},"第三，别复制粘贴。每节要有独立的信息，不是把几篇文章拼起来。",[15,288,289],{"id":289},"一个案例",[11,291,292],{},"2025 年 12 月，一个做数据库的客户，围绕\"数据库连接池\"聚合了 22 个长尾词，写了一篇 5000 字的指南。",[11,294,295],{},"结构是：连接池是什么（覆盖 5 个词）→ 怎么配置（6 个词）→ 常见报错（7 个词）→ 选型和监控（4 个词）。",[11,297,298],{},"上线三个月，22 个词里有 17 个进了前 20，9 个进前 10。单篇带来的月访问 4000 多。",[15,300,301],{"id":301},"一个提醒",[11,303,304],{},"聚合页写长，不代表可以不精。每个小节还是要回答清楚一个问题，读者能跳着看。",[11,306,307],{},"我们做聚合页的标准：随机抽一节给没看过的人看，他能看懂，就算合格。",[15,309,310],{"id":310},"一句判断",[11,312,313],{},"长尾聚合是\"用一篇好文章替代十篇平庸文章\"。前提是这篇真的好，而不是把十个词塞进一个页面。",[11,315,316],{},"写着写着发现写不深，就别聚合，拆成两篇反而更好。",{"title":89,"searchDepth":90,"depth":90,"links":318},[319,320,321,322,323,324,325],{"id":233,"depth":90,"text":233},{"id":242,"depth":90,"text":242},{"id":257,"depth":90,"text":258},{"id":273,"depth":90,"text":274},{"id":289,"depth":90,"text":289},{"id":301,"depth":90,"text":301},{"id":310,"depth":90,"text":310},"2026-08-23","长尾词一个个写文章成本太高，聚合到一个页面是常见做法。但聚合不等于把词堆进去。我们判断“该不该聚合”看三个条件，写法上分四层。一个页面覆盖 20 个词、排名进前 10 的案例也在这篇，照着结构写就行。",{},"\u002Fblog\u002Flongtail-aggregation",{"title":225,"description":327},"blog\u002Flongtail-aggregation",[333,113,334],"长尾词","关键词","Ti285lxOfB-eb9FIT2l7eL5Be7KsshIPW8FEF-bJfuo",1794088102588]