[{"data":1,"prerenderedAt":384},["ShallowReactive",2],{"blog-domestic-ai-citation-sources":3,"blog-prev-domestic-ai-citation-sources":217,"blog-next-domestic-ai-citation-sources":320},{"id":4,"title":5,"author":6,"body":7,"canonical":200,"category":201,"cover":200,"date":202,"description":203,"draft":204,"extension":205,"keywords":200,"meta":206,"navigation":207,"noindex":204,"ogImage":200,"path":208,"seo":209,"seoDescription":200,"seoTitle":200,"stem":210,"tags":211,"updated":200,"__hash__":216},"blog\u002Fblog\u002Fdomestic-ai-citation-sources.md","豆包、元宝、文心更爱引平台内容：一次 300 条引用的统计","元析智能",{"type":8,"value":9,"toc":191},"minimark",[10,14,18,21,24,124,127,130,137,143,149,152,155,158,161,164,167,170,173,176,179,182,185,188],[11,12,13],"p",{},"国内 AI 搜索引用官网的比例，比海外低得多。2026 年 3 月到 4 月，我们统计了豆包、元宝、文心三个引擎的 300 条引用，官网占 21%，剩余 79% 全在平台上。",[15,16,17],"h2",{"id":17},"统计怎么做的",[11,19,20],{},"三个引擎各跑 100 条查询，覆盖 B2B 软件方向的品类词、对比词、场景词。记录每条回答挂出来的来源，按域名分类，同一条回答里同一域名只算一次。",[11,22,23],{},"300 条引用，去重后 217 个来源域名。",[25,26,27,43],"table",{},[28,29,30],"thead",{},[31,32,33,37,40],"tr",{},[34,35,36],"th",{},"来源类型",[34,38,39],{},"占比",[34,41,42],{},"典型站点",[44,45,46,58,69,80,91,102,113],"tbody",{},[31,47,48,52,55],{},[49,50,51],"td",{},"官网",[49,53,54],{},"21%",[49,56,57],{},"客户自己的站",[31,59,60,63,66],{},[49,61,62],{},"问答",[49,64,65],{},"18%",[49,67,68],{},"知乎",[31,70,71,74,77],{},[49,72,73],{},"技术社区",[49,75,76],{},"14%",[49,78,79],{},"CSDN、掘金",[31,81,82,85,88],{},[49,83,84],{},"公众号转载",[49,86,87],{},"12%",[49,89,90],{},"各类转载站",[31,92,93,96,99],{},[49,94,95],{},"门户与媒体",[49,97,98],{},"11%",[49,100,101],{},"36 氪、少数派",[31,103,104,107,110],{},[49,105,106],{},"百科",[49,108,109],{},"8%",[49,111,112],{},"百度百科",[31,114,115,118,121],{},[49,116,117],{},"其他",[49,119,120],{},"16%",[49,122,123],{},"论坛、文档站、GitHub",[15,125,126],{"id":126},"为什么官网占比低",[11,128,129],{},"三个原因。",[11,131,132,136],{},[133,134,135],"strong",{},"平台有内容池。"," 国内引擎跟内容平台有合作，抓取通道和权重都不一样。同样一篇文章，发在平台上比发在自己站上更容易被读到。",[11,138,139,142],{},[133,140,141],{},"中文语料密度。"," 一个 B2B 软件官网通常几十个页面，知乎上关于这个品类的讨论可能是几千条。引擎在语料里找答案，谁多找谁。",[11,144,145,148],{},[133,146,147],{},"官网写得不像答案。"," 我们看的 100 个官网里，有 70 多个的页面结构是\"产品价值 + 功能列表 + 客户 logo\"。这种页面没有能独立回答问题的段落，引擎抓了也没法用。",[15,150,151],{"id":151},"公众号那块要注意",[11,153,154],{},"公众号占 12%，但要看清楚：引用的多是转载到可抓取页面的版本，公众号原文本身在很多引擎里抓不到。",[11,156,157],{},"这意味着两件事。一是发公众号的时候，同步发一份到官网或知乎，让内容有可抓的入口。二是看数据时别把\"公众号发了\"当成\"被收录了\"。",[11,159,160],{},"我们有个客户每月发 8 篇公众号，发了两年，国内 AI 回答里一次没出现过。查下来是内容从没在可抓页面出现过。",[15,162,163],{"id":163},"对软件公司的含义",[11,165,166],{},"官网是事实源，平台是入口。这个分工要想清楚。",[11,168,169],{},"官网负责\"准确\"：产品参数、定价、部署方式、公司信息，写清楚，让引擎有据可查。",[11,171,172],{},"平台负责\"被读到\"：知乎回答、技术社区文章、行业媒体稿，这些决定引擎会不会在回答里提你。",[11,174,175],{},"只做官网的客户，我们说句实话：国内这三个引擎里，你的可见度会一直很低，跟你内容质量没关系。",[15,177,178],{"id":178},"一个可执行的分配",[11,180,181],{},"按我们的经验，国内 GEO 的力气这么分：",[11,183,184],{},"官网占 40%，做事实和结构。平台占 50%，做分发和场景。剩下 10% 做监测。",[11,186,187],{},"不是五五开，是官网四、平台五。因为平台内容还承担了\"被别人引用\"的功能，一篇知乎回答可能同时被三个引擎读。",[11,189,190],{},"这个比例在海外反过来，官网大概占六。做国内和做出海，不能用同一套排期。",{"title":192,"searchDepth":193,"depth":193,"links":194},"",2,[195,196,197,198,199],{"id":17,"depth":193,"text":17},{"id":126,"depth":193,"text":126},{"id":151,"depth":193,"text":151},{"id":163,"depth":193,"text":163},{"id":178,"depth":193,"text":178},null,"行业观察","2026-03-23","统计国内三个引擎 300 条引用，官网只占 21%，知乎 18%、CSDN 14%、公众号 12%。国内做 GEO，官网是事实源，平台才是入口。",false,"md",{},true,"\u002Fblog\u002Fdomestic-ai-citation-sources",{"title":5,"description":203},"blog\u002Fdomestic-ai-citation-sources",[212,213,214,215],"国内AI搜索","豆包","元宝","平台内容","4xW_Ff2rK4r-EDskmVvW_gI45YGugT-ojM9hnwmABQ0",{"id":218,"title":219,"author":6,"body":220,"canonical":200,"category":308,"cover":200,"date":309,"description":310,"draft":204,"extension":205,"keywords":200,"meta":311,"navigation":207,"noindex":204,"ogImage":200,"path":312,"seo":313,"seoDescription":200,"seoTitle":200,"stem":314,"tags":315,"updated":200,"__hash__":319},"blog\u002Fblog\u002Fintent-mismatch-fix.md","意图错配让一个站两个月掉 34% 流量，我们这样修的",{"type":8,"value":221,"toc":298},[222,225,228,231,234,237,240,243,246,249,252,255,258,261,264,267,270,273,276,280,283,286,289,292,295],[11,223,224],{},"2026 年 1 月，一个客户找到我们，自然流量两个月掉了 34%。技术侧没问题，外链没少，文章也没删。查下来是 9 个页面承接了不属于它们的搜索意图。拆开之后六周，流量回到原来的 96%。",[15,226,227],{"id":227},"症状是什么样的",[11,229,230],{},"有曝光，没点击。或者点击了，十秒就跳出。搜索后台里，这些页面的曝光在涨，点击率却从 2.1% 掉到 0.8%。",[11,232,233],{},"排名也没有完全掉，很多词还在 8 到 15 名之间。这就是意图错配的典型信号：排得上，但用户不认。",[15,235,236],{"id":236},"怎么找出来",[11,238,239],{},"把查询级数据导出来，看每个页面承接了多少个不同意图的查询。一个页面同时接“是什么”“怎么选”“报错”三类词，基本就是错配。",[11,241,242],{},"我们那次找出 9 个页面，其中一个页面承接了 47 个查询，横跨三类意图。",[15,244,245],{"id":245},"为什么会变成这样",[11,247,248],{},"大多数是历史原因。页面一开始写的是学习类内容，后来运营为了省事，把对比信息和报错 FAQ 都堆进同一个页面，标题改过三次，正文越加越长。",[11,250,251],{},"页面变成了万能页，结果哪一类用户都不满意。",[15,253,254],{"id":254},"怎么修",[11,256,257],{},"能拆就拆。原来的页面留给最大的那类意图，另外两类各开一个新页面，内容重写，内链互相挂。",[11,259,260],{},"拆不了的，改页面类型。把博客文章改成文档页或对比页，标题、结构、CTA 全部重做。",[15,262,263],{"id":263},"修完的节奏",[11,265,266],{},"先改 3 个页面，观察四周，再改剩下 6 个。一次全改，出问题查不出原因。",[11,268,269],{},"那次我们第 2 周看到点击率回升，第 4 周排名稳住，第 6 周流量回到 96%。剩下的 4% 是季节波动，不是这次修改的问题。",[15,271,272],{"id":272},"怎么预防",[11,274,275],{},"词表里每个词标意图，页面立项时确认只接一种意图。这个规矩我们在 2025 年就写过，客户没执行，于是有了这次事故。",[15,277,279],{"id":278},"另一种错配页面没错标题错了","另一种错配：页面没错，标题错了",[11,281,282],{},"有些页面内容是对的，标题承诺了另一类意图。搜“怎么选”的人点进来，看到的是功能说明，一样会走。",[11,284,285],{},"这种情况改标题和首段就行，不用重写。我们那次 9 个页面里，有 3 个属于这一类，改完一周就看到点击率变化。",[15,287,288],{"id":288},"怎么防止再犯",[11,290,291],{},"新页面立项时写两行：目标词、目标意图。发布前由第二个人核对一遍，确认页面类型和意图匹配。",[11,293,294],{},"这个检查两分钟，能省掉后面几个月的返工。",[11,296,297],{},"意图错配不会立刻爆，它慢慢积累，等到某天算法调整或者竞品发力，才一次性显出来。",{"title":192,"searchDepth":193,"depth":193,"links":299},[300,301,302,303,304,305,306,307],{"id":227,"depth":193,"text":227},{"id":236,"depth":193,"text":236},{"id":245,"depth":193,"text":245},{"id":254,"depth":193,"text":254},{"id":263,"depth":193,"text":263},{"id":272,"depth":193,"text":272},{"id":278,"depth":193,"text":279},{"id":288,"depth":193,"text":288},"SEO","2026-03-22","2026 年 1 月一个客户自然流量两个月掉 34%，技术没动、外链没少、文章也没删。问题出在 9 个页面承接了不属于它的意图，拆开之后六周恢复。",{},"\u002Fblog\u002Fintent-mismatch-fix",{"title":219,"description":310},"blog\u002Fintent-mismatch-fix",[316,317,318],"搜索意图","流量下滑","复盘","UXVNi3Dk2O6JLxPKOT6zu6N1UmfpbmOw8nkSrWtPXNo",{"id":321,"title":322,"author":6,"body":323,"canonical":200,"category":373,"cover":200,"date":374,"description":375,"draft":204,"extension":205,"keywords":200,"meta":376,"navigation":207,"noindex":204,"ogImage":200,"path":377,"seo":378,"seoDescription":200,"seoTitle":200,"stem":379,"tags":380,"updated":200,"__hash__":383},"blog\u002Fblog\u002Fcontact-page-rebuild-review.md","联系页改版复盘：表单从九个字段砍到四个之后",{"type":8,"value":324,"toc":368},[325,328,332,335,338,341,344,347,350,353,356,359,362,365],[11,326,327],{},"2025 年 12 月我们给一个做流程管理的客户改联系页。改之前表单 9 个字段，提交率 2.1%；改之后 4 个字段，提交率 5.4%。但同一时期无效询盘从 12% 涨到 31%，这个副作用比提升更值得说。",[15,329,331],{"id":330},"改之前九个字段三屏","改之前：九个字段，三屏",[11,333,334],{},"原来的字段是：姓名、手机、邮箱、公司名、公司地址、行业、员工人数、预算范围、需求描述。移动端要滑三屏，最后一屏是\"需求描述\"的大文本框。页面上没有电话，没有微信，没有真人。",[11,336,337],{},"我们看了热力图，70% 的用户在第一屏之后离开，其中大部分停在\"公司地址\"那个字段。一个 B2B 站的联系页问地址，是想做线下拜访，但线上转化阶段问这个太早了。",[15,339,340],{"id":340},"改了什么",[11,342,343],{},"保留四个字段：姓名、手机、公司名、需求描述。手机保留是因为国内 B2B 必须电话跟进，但要加一句\"仅用于本次咨询\"。",[11,345,346],{},"删掉的五个字段不是不重要，是挪到后面。销售第一次通话时问行业和预算，比让用户自己填准确得多。",[11,348,349],{},"另外加了三个东西：一张负责人的工作照和名字，一个可预约的日历入口，一个可直接拨打的电话。这三样都是\"看得见的人\"，对转化帮助明显。",[15,351,352],{"id":352},"三周后的数据和副作用",[11,354,355],{},"提交率从 2.1% 涨到 5.4%，绝对数量从每周 9 个涨到 22 个。但无效询盘从 12% 涨到 31%，主要是同行调研和只想问价格的学生。",[11,357,358],{},"处理办法是在表单里加一个必选的\"你是\"下拉：甲方企业、渠道商、学生或个人、其他。选后两项的直接进另一个标签，销售不跟。加完这一项，无效询盘降到 19%，提交率只掉了 0.3 个百分点。",[11,360,361],{},"我们还试过把电话放在表单上方，拨打量一周涨了 14 个，但接起来一半是推销。后来把电话改成\"预约演示\"入口，接听质量明显好一些，虽然数量少了一半。",[11,363,364],{},"表单优化本质是转化问题，不是设计问题。字段删掉容易，删完怎么分层才是关键。",[11,366,367],{},"下一步我们准备把预约日历做成默认入口，表单退到第二选择。原因很简单：填表单是\"我留个联系方式等你来\"，预约是\"我选个时间我们聊\"，后者的意向度高一截。这个改动要等数据跑够两个月再定，急不得。",{"title":192,"searchDepth":193,"depth":193,"links":369},[370,371,372],{"id":330,"depth":193,"text":331},{"id":340,"depth":193,"text":340},{"id":352,"depth":193,"text":352},"内容运营","2026-03-24","2025 年 12 月我们给一个客户改联系页，字段从 9 个砍到 4 个，加了真人头像和预约入口。三周后提交率从 2.1% 涨到 5.4%，无效询盘也涨到 31%，这篇讲怎么处理这个副作用。",{},"\u002Fblog\u002Fcontact-page-rebuild-review",{"title":322,"description":375},"blog\u002Fcontact-page-rebuild-review",[381,382,318],"联系页","转化率","1C1IoQBS0NHXQurZqM2WzzzKcouVJBzLR_wktXI_Hdo",1794088107382]