[{"data":1,"prerenderedAt":419},["ShallowReactive",2],{"blog-monthly-content-review-metrics":3,"blog-prev-monthly-content-review-metrics":113,"blog-next-monthly-content-review-metrics":238},{"id":4,"title":5,"author":6,"body":7,"canonical":97,"category":98,"cover":97,"date":99,"description":100,"draft":101,"extension":102,"keywords":97,"meta":103,"navigation":104,"noindex":101,"ogImage":97,"path":105,"seo":106,"seoDescription":97,"seoTitle":97,"stem":107,"tags":108,"updated":97,"__hash__":112},"blog\u002Fblog\u002Fmonthly-content-review-metrics.md","月度内容复盘只看四个数，其他都是自娱自乐","元析智能",{"type":8,"value":9,"toc":85},"minimark",[10,14,18,21,24,27,30,33,36,39,42,45,48,51,55,58,61,64,67,70,73,76,79,82],[11,12,13],"p",{},"我们给客户做月度内容复盘，只放四个数：曝光、点击率、滚动到 75% 的比例、内容带来的注册或咨询。平均排名、字数、发文篇数都不进表。不是这些数没意义，是看了也做不出下一个动作。",[15,16,17],"h2",{"id":17},"四个数分别说明什么",[11,19,20],{},"曝光说明选题有没有被搜到，点击率说明标题和描述有没有让人想点，滚动深度说明内容有没有兑现点击，转化数说明这篇有没有商业价值。",[11,22,23],{},"四个数串起来就是一条链路，哪一段断了，下个月的活就往哪儿派。",[15,25,26],{"id":26},"平均排名为什么被砍掉",[11,28,29],{},"平均排名是按曝光加权摊平的一个数。加了 200 个长尾曝光，平均排名会变好看，但一篇都没进前 20。",[11,31,32],{},"2025 年 10 月我们给一个客户做复盘，平均排名从 18 涨到 14，客户挺高兴。我翻到查询级数据，发现进前 10 的查询只多了 2 个。那个月真正变好的是曝光，排名只是被平均出来的。",[15,34,35],{"id":35},"点击率是最被低估的数",[11,37,38],{},"同样排第 5，点击率 1.5% 和 3% 是两倍流量差距。改标题和描述的成本，比写一篇新文章低十倍。",[11,40,41],{},"2025 年 11 月我们给一个开发者工具站改了 12 个页面的标题和描述，只调整了把结论提前和加上数字，一个月后这些页面点击率从 1.8% 到 3.1%，曝光基本没变。",[15,43,44],{"id":44},"转化数怎么归因才不算错",[11,46,47],{},"最后点击归因会把内容的价值算给品牌词和首页，内容看起来毫无贡献。我们的做法是加一个“内容首触”标记：用户第一次进来是内容页，90 天内成交都算这篇内容一份。",[11,49,50],{},"这个标记要在 GA4 里配，配一次能用一年，比每月解释数据便宜。",[15,52,54],{"id":53},"复盘会开-30-分钟","复盘会开 30 分钟",[11,56,57],{},"前 10 分钟看四个数，中间 10 分钟挑出两个变差的页面，最后 10 分钟定下个月的三个动作。定不出动作的复盘就是汇报，不叫复盘。",[15,59,60],{"id":60},"还有一个数我们单独看",[11,62,63],{},"AI 引用次数。每月手工查 20 个目标问题，记录有没有被引用。这个数不进月度表，季度看一次，因为波动太大。",[15,65,66],{"id":66},"数据不对齐怎么处理",[11,68,69],{},"客户的 GA4、搜索后台、表单后台三个数经常差 20% 以上。别在复盘会上解释差异，先把口径固定下来：曝光和点击只认搜索后台，转化只认表单和 CRM。",[11,71,72],{},"差异留给月报附录，一句话说明。把时间花在解释差异上，复盘会就变成了对账会。",[15,74,75],{"id":75},"一个我们停掉的动作",[11,77,78],{},"以前每月会统计“发了多少篇”“总字数多少”。这两个数除了自我安慰没有用途，2025 年 6 月停掉之后，复盘会省了 15 分钟。",[11,80,81],{},"填表的人也不必为了好看硬凑篇数，内容质量反而稳了。",[11,83,84],{},"复盘不是汇报，是决定下个月写什么、改什么、停什么。",{"title":86,"searchDepth":87,"depth":87,"links":88},"",2,[89,90,91,92,93,94,95,96],{"id":17,"depth":87,"text":17},{"id":26,"depth":87,"text":26},{"id":35,"depth":87,"text":35},{"id":44,"depth":87,"text":44},{"id":53,"depth":87,"text":54},{"id":60,"depth":87,"text":60},{"id":66,"depth":87,"text":66},{"id":75,"depth":87,"text":75},null,"内容运营","2025-12-15","我们的月度复盘只有四个指标：曝光、点击率、滚动到 75% 的比例、内容带来的注册或咨询。平均排名和字数早就不看了，看了也做不出动作。",false,"md",{},true,"\u002Fblog\u002Fmonthly-content-review-metrics",{"title":5,"description":100},"blog\u002Fmonthly-content-review-metrics",[109,110,111],"内容评估","数据复盘","SEO指标","Xsg0ZzS3dnESC3_wapTngzpCkf-dlFUJSmXS2FpeNWs",{"id":114,"title":115,"author":6,"body":116,"canonical":97,"category":226,"cover":97,"date":227,"description":228,"draft":101,"extension":102,"keywords":97,"meta":229,"navigation":104,"noindex":101,"ogImage":97,"path":230,"seo":231,"seoDescription":97,"seoTitle":97,"stem":232,"tags":233,"updated":97,"__hash__":237},"blog\u002Fblog\u002Fttfb-network-vs-server.md","TTFB 高先分清是网络还是服务端，别急着换机房",{"type":8,"value":117,"toc":219},[118,121,124,162,168,172,175,178,182,185,188,191,194,197,200,203,206,209,212,215],[11,119,120],{},"TTFB 高的时候，先别换机房。今年 1 月我们查过一个 SaaS 的首页，移动端首字节 1.4 秒，团队第一反应是\"服务器不行\"，准备换云厂商。查下来 800 毫秒花在服务端自己身上，换到哪儿都一样。",[11,122,123],{},"先把 TTFB 拆开看。用 curl 的耗时分解，一个请求能拆成五段：",[125,126,130],"pre",{"className":127,"code":128,"language":129,"meta":86,"style":86},"language-bash shiki shiki-themes github-light github-dark","curl -o \u002Fdev\u002Fnull -s -w \"dns:%{time_namelookup} tcp:%{time_connect} tls:%{time_appconnect} ttfb:%{time_starttransfer}\\n\" https:\u002F\u002Fexample.com\u002F\n","bash",[131,132,133],"code",{"__ignoreMap":86},[134,135,138,142,146,150,153,156,159],"span",{"class":136,"line":137},"line",1,[134,139,141],{"class":140},"sScJk","curl",[134,143,145],{"class":144},"sj4cs"," -o",[134,147,149],{"class":148},"sZZnC"," \u002Fdev\u002Fnull",[134,151,152],{"class":144}," -s",[134,154,155],{"class":144}," -w",[134,157,158],{"class":148}," \"dns:%{time_namelookup} tcp:%{time_connect} tls:%{time_appconnect} ttfb:%{time_starttransfer}\\n\"",[134,160,161],{"class":148}," https:\u002F\u002Fexample.com\u002F\n",[11,163,164,167],{},[131,165,166],{},"time_starttransfer"," 减去前面几段，剩下的才是服务端处理时间。这个站 DNS 30 毫秒、TLS 120 毫秒、传输 90 毫秒，剩下 800 毫秒就是后端在算。",[15,169,171],{"id":170},"服务端那-800-毫秒在干什么","服务端那 800 毫秒在干什么",[11,173,174],{},"那个站一次页面请求串行调了 6 个接口，最慢的一个 340 毫秒。6 个加起来 1.1 秒，有并发，实际 800 毫秒。前端一直以为是网络问题，因为监控上只看到\"页面慢\"三个字。",[11,176,177],{},"查法是看应用侧的链路追踪，或者最土的办法：在服务端入口加时间日志，记每个接口的耗时。这个客户加完日志第二天就定位了，比讨论一周有用。",[15,179,181],{"id":180},"改法能并行的并行能缓存的缓存","改法：能并行的并行，能缓存的缓存",[11,183,184],{},"我们做了两件事：6 个接口里 4 个改成并行，总耗时从 800 降到 260 毫秒；首页那块\"最新公告\"从实时查库改成 60 秒缓存。TTFB 从 1.4 秒降到 480 毫秒，同一批机器。",[15,186,187],{"id":187},"什么情况才该怀疑网络",[11,189,190],{},"三种情况我才会建议查链路：TTFB 在 200 到 400 毫秒之间但各地差异大；静态资源也慢；TLS 握手占 300 毫秒以上。这三种通常指向线路或者 CDN 节点问题，和服务端代码无关。",[11,192,193],{},"还有个粗糙但好用的判断：如果你的静态 HTML 的 TTFB 就很高，那是基础设施；如果只有动态页慢，那基本是代码。这个判断一分钟能做完，比开会讨论有用。",[15,195,196],{"id":196},"一个容易误判的情况",[11,198,199],{},"首页慢、产品页快，这种情况很常见，因为首页的接口最多。我们查过一个站，产品页 TTFB 210 毫秒，首页 1.3 秒，团队按首页的平均值去投诉机房，方向完全错了。",[11,201,202],{},"正确做法是按页面类型分组统计 TTFB，每种取 P75。营销页、文档页、后台页分开看，差异往往比你想的大。",[15,204,205],{"id":205},"我们给客户的检查顺序",[11,207,208],{},"第一步，挑三个页面（首页、一个产品页、一个文档页）用 curl 各跑五次，记下五段耗时。第二步，把服务端那一段和接口日志对上。第三步，才决定是改代码、加缓存还是查线路。",[11,210,211],{},"这个顺序的好处是成本低。三步做完，半小时，能排除八成误判。最怕的是先入为主地认为是网络问题，换了一圈机房，最后发现是接口没并行。",[11,213,214],{},"最后一句：换机房是成本最高、见效最不确定的一招。我们接手过的性能项目里，真正靠换机房解决问题的只有两次，都是国内单线机房那种情况。",[216,217,218],"style",{},"html pre.shiki code .sScJk, html code.shiki .sScJk{--shiki-default:#6F42C1;--shiki-dark:#B392F0}html pre.shiki code .sj4cs, html code.shiki .sj4cs{--shiki-default:#005CC5;--shiki-dark:#79B8FF}html pre.shiki code .sZZnC, html code.shiki .sZZnC{--shiki-default:#032F62;--shiki-dark:#9ECBFF}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"title":86,"searchDepth":87,"depth":87,"links":220},[221,222,223,224,225],{"id":170,"depth":87,"text":171},{"id":180,"depth":87,"text":181},{"id":187,"depth":87,"text":187},{"id":196,"depth":87,"text":196},{"id":205,"depth":87,"text":205},"技术SEO","2025-12-14","一个 SaaS 首页移动端首字节 1.4 秒，团队准备换云厂商，其实 800 毫秒花在服务端串行接口上。给一条 curl 拆时间的命令，和三个才该怀疑网络的情况。",{},"\u002Fblog\u002Fttfb-network-vs-server",{"title":115,"description":228},"blog\u002Fttfb-network-vs-server",[234,235,236],"TTFB","服务端性能","性能诊断","BR70gBacE8JBbLjWfYgmkRwPIixD9DwNRj0tqZXvyIE",{"id":239,"title":240,"author":6,"body":241,"canonical":97,"category":406,"cover":97,"date":407,"description":408,"draft":101,"extension":102,"keywords":97,"meta":409,"navigation":104,"noindex":101,"ogImage":97,"path":410,"seo":411,"seoDescription":97,"seoTitle":97,"stem":412,"tags":413,"updated":97,"__hash__":418},"blog\u002Fblog\u002Fcompetitor-ai-share-audit.md","竞品在 AI 回答里占多少位：一次 320 条回答的人工盘点",{"type":8,"value":242,"toc":399},[243,246,249,252,255,259,262,269,275,281,287,290,293,296,299,302,305,311,317,323,326,393,396],[11,244,245],{},"先给参照系，再谈优化。2025 年 12 月我们给 6 个 B2B 软件客户各跑了 80 个查询，四个 AI 引擎，记录每条回答里出现了哪些品牌。结果很一致：竞品平均出现在 43% 的回答里，客户自己 19%。",[15,247,248],{"id":248},"为什么先做这件事",[11,250,251],{},"老板问\"我们在 AI 里有没有存在感\"，你答\"有一些\"，这对话没法继续。换成\"竞品 A 在 51 个回答里出现，我们 15 个，其中 6 个是竞品 A 被列为推荐、我们被列为备选\"，讨论立刻具体。",[11,253,254],{},"参照系还有个作用：它挡掉无效焦虑。有些客户看到自己只被提 10 次就想大改官网，一查竞品也就 12 次——这个品类在 AI 里的整体水位就低，不需要动内容，需要等语料积累。",[15,256,258],{"id":257},"_80-个查询怎么选","80 个查询怎么选",[11,260,261],{},"四类，每类 20 个：",[11,263,264,268],{},[265,266,267],"strong",{},"品类词。","\"数据同步工具\"\"API 网关\"\"日志分析平台\"。这类问的是\"有什么\"，AI 会列举。",[11,270,271,274],{},[265,272,273],{},"对比词。","\"A 和 B 的区别\"\"A 的替代品\"。这类问的是\"选哪个\"，AI 会给判断。",[11,276,277,280],{},[265,278,279],{},"场景词。","\"MySQL 到 ClickHouse 怎么做增量同步\"\"K8s 集群怎么做鉴权\"。这类问的是\"怎么做\"，AI 会引文档。",[11,282,283,286],{},[265,284,285],{},"品牌加场景。","\"XX 能接 Kafka 吗\"\"XX 支持私有部署吗\"。这类问的是\"你能不能\"，AI 会引官网和第三方。",[11,288,289],{},"品类词和对比词决定获客，场景词决定你是不是内行，品牌加场景决定成单前的最后一步。四类都跑，才知道短板在哪。",[15,291,292],{"id":292},"记录哪四列",[11,294,295],{},"一列查询，一列引擎，一列\"出现方式\"，一列原文。",[11,297,298],{},"出现方式分五种：首位推荐、列举中的一项、被当备选、被负面提及、没出现。这五种的商业价值差很远。首位推荐和列举中一项，看着都是\"出现了\"，实际差一个量级。",[11,300,301],{},"原文必须抄下来。只记\"出现了\"没有用，要看它用什么话描述你。我们有个客户被描述成\"轻量级方案\"，一看就知道 AI 是从他们 2023 年的旧文档里学的。",[15,303,304],{"id":304},"三个高频发现",[11,306,307,310],{},[265,308,309],{},"竞品在\"替代品\"类查询里占位最稳。"," 这类查询我们跑了 6 个客户，竞品平均出现率 61%。原因是替代品文章是最好写的选题，第三方媒体爱写，AI 爱引。",[11,312,313,316],{},[265,314,315],{},"客户在\"哪个好\"里普遍缺席。"," 80 个查询里，客户在对比类回答中出现 9 次。对比类内容要么自己写、要么被别人写，不写就是空白。",[11,318,319,322],{},[265,320,321],{},"出现方式比出现次数重要。"," 有个客户出现 22 次，其中 17 次是\"也有人用\"这种备选位，实际带来 0 个询盘。另一个客户出现 11 次，8 次在首位，一个月 6 个询盘。",[15,324,325],{"id":325},"一张表长什么样",[327,328,329,348],"table",{},[330,331,332],"thead",{},[333,334,335,339,342,345],"tr",{},[336,337,338],"th",{},"查询",[336,340,341],{},"引擎",[336,343,344],{},"出现方式",[336,346,347],{},"提到的品牌",[349,350,351,366,380],"tbody",{},[333,352,353,357,360,363],{},[354,355,356],"td",{},"数据同步工具推荐",[354,358,359],{},"ChatGPT",[354,361,362],{},"列举第 4 项",[354,364,365],{},"竞品 A、竞品 B、客户",[333,367,368,371,374,377],{},[354,369,370],{},"XX 的替代品",[354,372,373],{},"Perplexity",[354,375,376],{},"未出现",[354,378,379],{},"竞品 A、竞品 C",[333,381,382,385,387,390],{},[354,383,384],{},"Kafka 增量同步怎么做",[354,386,359],{},[354,388,389],{},"引官网文档",[354,391,392],{},"客户",[11,394,395],{},"这张表跑一轮 4 小时，比任何工具的报告都好用，因为你能读到原文。",[11,397,398],{},"先知道自己站在哪，再决定往哪走。",{"title":86,"searchDepth":87,"depth":87,"links":400},[401,402,403,404,405],{"id":248,"depth":87,"text":248},{"id":257,"depth":87,"text":258},{"id":292,"depth":87,"text":292},{"id":304,"depth":87,"text":304},{"id":325,"depth":87,"text":325},"行业观察","2025-12-16","给 6 个客户跑了 80 个查询，竞品平均出现在 43% 的 AI 回答里，客户自己只有 19%。先建参照系再谈优化，附查询分类和记录表。",{},"\u002Fblog\u002Fcompetitor-ai-share-audit",{"title":240,"description":408},"blog\u002Fcompetitor-ai-share-audit",[414,415,416,417],"GEO","竞品分析","AI搜索","可见性监测","pz-2xLuhs9v9BEPytMg8XH3IdiR8jSkmL8DTjHLA1ws",1794088110019]