[{"data":1,"prerenderedAt":741},["ShallowReactive",2],{"blog-serp-reputation-repair-order":3,"blog-prev-serp-reputation-repair-order":92,"blog-next-serp-reputation-repair-order":229},{"id":4,"title":5,"author":6,"body":7,"canonical":76,"category":77,"cover":76,"date":78,"description":79,"draft":80,"extension":81,"keywords":76,"meta":82,"navigation":83,"noindex":80,"ogImage":76,"path":84,"seo":85,"seoDescription":76,"seoTitle":76,"stem":86,"tags":87,"updated":76,"__hash__":91},"blog\u002Fblog\u002Fserp-reputation-repair-order.md","品牌搜索结果首页全是负面，先别急着找删帖","元析智能",{"type":8,"value":9,"toc":69},"minimark",[10,14,19,22,25,28,32,39,45,51,57,60,63,66],[11,12,13],"p",{},"搜公司名，第一页一半是负面，老板第一反应是\"找人删掉\"。这个反应是错的，大部分负面删不掉，而且顺序错了会把事情弄大。先判断性质，再决定动作，这是我们处理的固定顺序。",[15,16,18],"h2",{"id":17},"哪些能删哪些删不掉","哪些能删，哪些删不掉",[11,20,21],{},"能处理的是三类。一是事实错误或造谣，走法务和平台投诉，成功率取决于证据。二是平台违规内容，比如人身攻击、泄露隐私，走平台举报通道。三是你自己站上的旧页面，改标题、加声明或者直接下线，这个最快。",[11,23,24],{},"删不掉的是三类。真实的用户差评，只要内容合规，平台不会删。媒体的报道，除非能证明事实错误，否则只能要求补充回应。论坛和问答里的讨论，删了一个帖子会有十个新帖。",[11,26,27],{},"分清这两组，是省钱的第一步。找错方向，花几万块也是打水漂。",[15,29,31],{"id":30},"处理顺序问答式说清","处理顺序，问答式说清",[11,33,34,38],{},[35,36,37],"strong",{},"问：第一步做什么？"," 先截图存证，把负面按性质分类，再看它排在什么位置、有没有被AI搜索引用。事实错误的走投诉，真实差评走回应与占位，别混在一起谈。",[11,40,41,44],{},[35,42,43],{},"问：投诉多久见效？"," 平台投诉通常 7 到 30 天有结果，法务路径按案件走，可能几个月。要做时间预期，别以为发个函明天就没了。",[11,46,47,50],{},[35,48,49],{},"问：能不能找\"删帖公司\"？"," 不建议。付费删帖本身有法律风险，而且这个领域骗子多，收钱不办事是常态。我们见过客户付了两万八，帖子还在，对方失联。",[11,52,53,56],{},[35,54,55],{},"问：多久能改善搜索结果？"," 靠内容占位，3 到 6 个月能看到变化。前提是内容本身有排名能力，光发几篇通稿堆不上去。",[15,58,59],{"id":59},"我们实际在做的事",[11,61,62],{},"一页 10 条结果，我们争取占 5 到 6 条：官网首页、关于页、案例页、两三个第三方平台的正面内容、一条媒体报道。占住这些位置，负面自然被挤到第二页，效果比删帖稳，也不会反弹。",[11,64,65],{},"2025 年 9 月我们处理过一个客户，第一页 4 条负面，其中 2 条是真实差评。我们没有碰差评，做了三件事：把关于页和案例页的标题改得更贴近品牌词，补了两篇回应式的说明文章，推动两个合作方在自家站点提到客户。四个月后，第一页还剩 1 条负面，排在第八位。",[11,67,68],{},"能删的很少，能盖的很多。但这个\"盖\"不是刷水军，是让自己的官方信息在搜索结果里足够多、足够清楚。",{"title":70,"searchDepth":71,"depth":71,"links":72},"",2,[73,74,75],{"id":17,"depth":71,"text":18},{"id":30,"depth":71,"text":31},{"id":59,"depth":71,"text":59},null,"行业观察","2026-01-04","负面信息处理的顺序错了，钱就白花。这篇用问答讲清楚：能删的只有哪三类、先做什么后做什么、为什么\"占位\"比\"删帖\"现实，以及踩过的坑。",false,"md",{},true,"\u002Fblog\u002Fserp-reputation-repair-order",{"title":5,"description":79},"blog\u002Fserp-reputation-repair-order",[88,89,90],"品牌口碑","负面处理","搜索结果","qunjBXcHPzN3h5G2LZXPbFruGaRHmllct6FcTNvAxis",{"id":93,"title":94,"author":6,"body":95,"canonical":76,"category":216,"cover":76,"date":217,"description":218,"draft":80,"extension":81,"keywords":76,"meta":219,"navigation":83,"noindex":80,"ogImage":76,"path":220,"seo":221,"seoDescription":76,"seoTitle":76,"stem":222,"tags":223,"updated":76,"__hash__":228},"blog\u002Fblog\u002Fai-citation-rate-denominator.md","AI 引用率怎么算：分母选错，同一个团队能得出两个结论",{"type":8,"value":96,"toc":208},[97,100,104,110,116,122,125,128,131,134,137,140,143,156,159,162,165,168,171,174,177,180,190,193,196,199,202,205],[11,98,99],{},"AI 引用率这个数，八成算错在分母。同一批数据，用品牌词做分母能算出 23%，换成品类词只剩 9%。差 14 个百分点，足够让一个团队做错半年的决定。",[15,101,103],{"id":102},"三种分母三个结论","三种分母，三个结论",[11,105,106,109],{},[35,107,108],{},"分母是品牌词。"," 用户搜的就是你，AI 回答里出现你的概率自然高。测出来通常 20% 以上，好看，没用。这个数只说明你还没被 AI 忘掉。",[11,111,112,115],{},[35,113,114],{},"分母是品类词。"," 用户搜\"数据同步工具\"，没指定谁。测出来通常 5% 到 15%。这才是获客相关的口径。",[11,117,118,121],{},[35,119,120],{},"分母是全量查询。"," 把你后台所有关键词都算进去，包括那些你根本不该出现的词。测出来接近 0，除了打击士气没别的用。",[11,123,124],{},"我们给客户的标准是第二种：品类词加对比词，固定 60 个。",[15,126,127],{"id":127},"我们的固定口径",[11,129,130],{},"60 个查询，四类各 15 个：品类词、对比词、场景词、品牌加场景。",[11,132,133],{},"每月同一批、同一时间、同样的引擎列表。不换词，不加减。听着死板，但只有固定才能比。",[11,135,136],{},"有个客户中途要求加 20 个新词，说新品类要测。我们单开一张表记，不混进主口径。混进去之后环比就没意义了。",[15,138,139],{"id":139},"分子怎么算",[11,141,142],{},"三种出现方式，权重不同：",[144,145,146,150,153],"ul",{},[147,148,149],"li",{},"品牌名或产品名出现，计 1 分",[147,151,152],{},"出现官网链接，计 2 分",[147,154,155],{},"出现在首位推荐或作为主答案，计 3 分",[11,157,158],{},"单条回答取最高分，不叠加。60 个查询跑 4 个引擎，满分 720 分。",[11,160,161],{},"这不是精确科学，是把\"出现了\"和\"被推荐了\"分开。我们试过只记二元（出现\u002F没出现），结果一个客户的分数涨了 30%，实际全是\"也有人用\"这类备选位，询盘没动。",[15,163,164],{"id":164},"一个被虚高的例子",[11,166,167],{},"2025 年 12 月，一个客户自己算的引用率是 23%，拿来跟我们说做得不错。",[11,169,170],{},"我们按品类词重算，9%。差异来自哪儿？他们的品牌词占了查询池的 60%。也就是说，他们一直在测\"AI 认不认识我\"，而不是\"AI 会不会推荐我\"。",[11,172,173],{},"重算之后，发现真正的问题是他们在对比类查询里几乎不出现。改的方向从\"多发品牌新闻\"变成了\"补对比页和第三方评测\"。",[15,175,176],{"id":176},"给老板怎么讲",[11,178,179],{},"别用一个数字。给两个：",[11,181,182,185,186,189],{},[35,183,184],{},"覆盖率","，60 个查询里有多少条回答提到你，反映存在感。\n",[35,187,188],{},"首位率","，提到你的那些里，多少在首位，反映推荐强度。",[11,191,192],{},"覆盖率 40%、首位率 10%，说明大家知道你但没人首选你。覆盖率 12%、首位率 60%，说明小众但精准。",[11,194,195],{},"两个数一起看，才不会互相骗。",[15,197,198],{"id":198},"口径要写进报告首页",[11,200,201],{},"我们现在给客户的报告第一页固定三行：查询池是哪 60 个、引擎是哪 4 个、每月的采集日期。谁都能照着复现。",[11,203,204],{},"这看着是形式主义，实际能省很多吵架。有次客户换了市场负责人，新来的第一句话是\"这个数怎么算的\"。翻到第一页，三行字解释清楚，会议十分钟结束。",[11,206,207],{},"反过来，口径不写的报告，数字一变就会被怀疑在挑数据。GEO 这行现在没有第三方审计，可信度只能靠口径透明撑着。",{"title":70,"searchDepth":71,"depth":71,"links":209},[210,211,212,213,214,215],{"id":102,"depth":71,"text":103},{"id":127,"depth":71,"text":127},{"id":139,"depth":71,"text":139},{"id":164,"depth":71,"text":164},{"id":176,"depth":71,"text":176},{"id":198,"depth":71,"text":198},"工具评测","2026-01-03","同一份数据，用品牌词做分母算出 23%，换成品类词只有 9%。分母决定结论。我们固定 60 个品类词和对比词，每月同一批，口径写进报告。",{},"\u002Fblog\u002Fai-citation-rate-denominator",{"title":94,"description":218},"blog\u002Fai-citation-rate-denominator",[224,225,226,227],"AI引用率","监测方法","GEO","数据口径","RE-y7EH5R6F86YHzfu_UR0sJCJhJUPLhkgKVRm1Ttz8",{"id":230,"title":231,"author":6,"body":232,"canonical":76,"category":226,"cover":76,"date":730,"description":731,"draft":80,"extension":81,"keywords":76,"meta":732,"navigation":83,"noindex":80,"ogImage":76,"path":733,"seo":734,"seoDescription":76,"seoTitle":76,"stem":735,"tags":736,"updated":76,"__hash__":740},"blog\u002Fblog\u002Fgeo-content-structure.md","让 AI 引擎引用你的 6 个内容结构技巧",{"type":8,"value":233,"toc":720},[234,237,241,244,250,256,261,266,269,274,285,289,292,306,311,316,321,326,333,336,340,354,358,361,365,370,374,379,399,402,431,434,438,441,446,456,459,464,470,477,481,495,499,502,507,512,515,520,528,532,546,550,553,559,564,569,574,640,643,649,654,665,670,681,684,687,708,711,714,717],[11,235,236],{},"GEO 做了半年,我们发现 AI 引擎引用内容有明确偏好。不是内容好就行,是内容结构要符合 AI 抓取的逻辑。这篇文章写我们实测有效的 6 个内容结构技巧,每个配 before\u002Fafter 例子。",[15,238,240],{"id":239},"技巧-1结论前置","技巧 1:结论前置",[11,242,243],{},"AI 抓摘要,前 100 字没有明确结论,基本不会被引用。结论要直接、要带立场,不要铺垫。",[11,245,246,249],{},[35,247,248],{},"Before(不会被引用)",":",[251,252,253],"blockquote",{},[11,254,255],{},"在数字化浪潮中,SEO 作为一项重要的网络营销手段,其重要性日益凸显。许多企业都在思考如何通过 SEO 来提升品牌曝光度和市场竞争力。本文将从多个角度探讨 SEO 的价值和实施方法。",[11,257,258,249],{},[35,259,260],{},"After(会被引用)",[251,262,263],{},[11,264,265],{},"2025 年百度算法更新后,低质外链权重下降 60%。如果你最近流量下滑,先查外链。",[11,267,268],{},"判断标准:把前 100 字抽出来,能不能独立回答用户问题?能 = 合格。不能 = 重写。",[11,270,271,249],{},[35,272,273],{},"实操要点",[144,275,276,279,282],{},[147,277,278],{},"第一句给结论,不要\"随着...越来越...\"开头",[147,280,281],{},"前两段不要绕,直接给判断和数据",[147,283,284],{},"结论要带立场,AI 不引用骑墙表达",[15,286,288],{"id":287},"技巧-2可引用的事实句","技巧 2:可引用的事实句",[11,290,291],{},"AI 偏好明确陈述句,不引用模糊表达。判断一个句子能不能被引用:",[144,293,294,300],{},[147,295,296,299],{},[35,297,298],{},"可引用",":有具体数字、时间、主体、动作",[147,301,302,305],{},[35,303,304],{},"不可引用",":有\"可能\"\"某种程度上\"\"往往\"\"通常\"",[11,307,308,249],{},[35,309,310],{},"Before",[251,312,313],{},[11,314,315],{},"外链权重可能在某种程度上有所变化,具体影响因站点而异。",[11,317,318,249],{},[35,319,320],{},"After",[251,322,323],{},[11,324,325],{},"百度 2025 年 3 月搜索质量报告显示,低质外链权重下降 60%。",[11,327,328,329,332],{},"可引用句的公式:",[35,330,331],{},"主体 + 时间 + 动作 + 具体数字 + 出处","。",[11,334,335],{},"我们做 GEO 时,每篇内容至少包含 3-5 个可引用事实句,放在段落开头或独立成段。AI 抓取时会优先选这种句子。",[11,337,338,249],{},[35,339,273],{},[144,341,342,345,348,351],{},[147,343,344],{},"数字要具体,不要\"大量\"\"许多\"",[147,346,347],{},"时间要明确,不要\"近期\"\"近年来\"",[147,349,350],{},"出处要可验证,不要\"专家表示\"",[147,352,353],{},"主体要清楚,不要\"业界认为\"",[15,355,357],{"id":356},"技巧-3数据带出处的格式","技巧 3:数据带出处的格式",[11,359,360],{},"AI 引用需要可信源。带出处和年份的数据,被引用概率比无出处数据高 3-5 倍(我们自己监测的)。",[11,362,363,249],{},[35,364,310],{},[251,366,367],{},[11,368,369],{},"研究表明,首屏搜索结果的平均外链数远高于其他位置。",[11,371,372,249],{},[35,373,320],{},[251,375,376],{},[11,377,378],{},"Backlinko 2024 年分析 120 万个搜索结果,发现首屏结果平均外链数是第 10 名的 3.8 倍。",[11,380,381,382,332],{},"格式模板:",[35,383,384,388,389,388,392,395,396],{},[385,386,387],"span",{},"来源"," ",[385,390,391],{},"年份",[385,393,394],{},"研究方法\u002F样本量",",发现 ",[385,397,398],{},"具体数据",[11,400,401],{},"注意几个细节:",[144,403,404,413,419,425],{},[147,405,406,409,412],{},[35,407,408],{},"出处放句首",[410,411],"backlinko",{}," 2024 年分析...,不是\"...根据 Backlinko 的研究\"",[147,414,415,418],{},[35,416,417],{},"样本量要写",":\"分析 120 万个搜索结果\"比\"研究显示\"可信",[147,420,421,424],{},[35,422,423],{},"数字要精确",":3.8 倍,不是\"几倍\"",[147,426,427,430],{},[35,428,429],{},"来源要可查",":知名研究机构、官方数据、权威媒体",[11,432,433],{},"我们监测发现,AI 引擎对 Backlinko、Ahrefs、Statista、SimilarWeb 这些来源的数据引用率高。模糊来源(没有具体出处的\"研究表明\")基本不引用。",[15,435,437],{"id":436},"技巧-4faq-结构怎么写最容易被抓","技巧 4:FAQ 结构怎么写最容易被抓",[11,439,440],{},"FAQ 是 GEO 里最容易出效果的格式。但 FAQ 怎么写有讲究。",[11,442,443,249],{},[35,444,445],{},"Before(不容易抓)",[447,448,453],"pre",{"className":449,"code":451,"language":452},[450],"language-text","常见问题\nQ1: SEO 多久见效?\nA1: SEO 见效时间因多种因素而异,包括网站基础、竞争程度、投入资源等。一般来说,新站需要 6-12 个月,老站需要 3-6 个月。具体请咨询专业团队。\n\nQ2: GEO 是什么?\nA2: GEO 是一种新兴的搜索引擎优化方式,旨在提升内容在 AI 搜索引擎中的可见度...\n","text",[454,455,451],"code",{"__ignoreMap":70},[11,457,458],{},"问题:答案太啰嗦,套话多,没有具体数据。AI 不会抓这种内容。",[11,460,461,249],{},[35,462,463],{},"After(容易被抓)",[447,465,468],{"className":466,"code":467,"language":452},[450],"Q: GEO 多久见效?\nA: 比 SEO 快一些。内容结构化后,2-4 周可能就开始被引用。稳定引用需要 2-4 个月持续建设。\n\nQ: GEO 和 SEO 冲突吗?\nA: 不冲突,互补。SEO 打搜索流量,GEO 打 AI 引用流量,建议两个一起做。\n",[454,469,467],{"__ignoreMap":70},[11,471,472,473,476],{},"判断标准:",[35,474,475],{},"答案能不能 50 字内说完?"," 能 = 合格。不能 = 答案太啰嗦。",[11,478,479,249],{},[35,480,273],{},[144,482,483,486,489,492],{},[147,484,485],{},"Q 用用户问法,不要改写成书面语(\"GEO 多久见效\"比\"关于 GEO 见效周期的探讨\"好)",[147,487,488],{},"A 第一句给结论,后面最多补 1-2 句展开",[147,490,491],{},"答案控制在 30-80 字,不要长篇",[147,493,494],{},"每个答案带 1 个具体数字或时间",[15,496,498],{"id":497},"技巧-5段落主题句原则","技巧 5:段落主题句原则",[11,500,501],{},"AI 按段落抓取。一段一个主题,首句即主题句,后面是展开。",[11,503,504,249],{},[35,505,506],{},"Before(抓不到主题)",[251,508,509],{},[11,510,511],{},"SEO 和 GEO 虽然有区别,但也有很多共同点。SEO 注重关键词和外链,GEO 注重内容结构和权威度。不过在技术层面,两者都需要保证网站的可抓取性。此外,内容质量都是基础...",[11,513,514],{},"问题:一段讲三件事,AI 不知道这段在讲什么。",[11,516,517,249],{},[35,518,519],{},"After(主题清晰)",[251,521,522,525],{},[11,523,524],{},"SEO 和 GEO 在技术层面是相通的。两者都需要保证网站可抓取、内容可索引、结构化数据规范。技术 SEO 做好,GEO 也直接受益。",[11,526,527],{},"内容策略上两者开始分化。SEO 注重关键词覆盖和外链建设,GEO 注重内容结构和可引用性。同样一篇内容,SEO 视角看关键词密度,GEO 视角看事实句密度。",[11,529,530,249],{},[35,531,273],{},[144,533,534,537,540,543],{},[147,535,536],{},"一段只讲一件事",[147,538,539],{},"首句是主题句,概括这段主旨",[147,541,542],{},"段落长度 50-150 字,不要太长",[147,544,545],{},"段落之间用空行分隔",[15,547,549],{"id":548},"技巧-6表格和列表对-ai-抓取的影响","技巧 6:表格和列表对 AI 抓取的影响",[11,551,552],{},"很多人不知道,表格和列表对 AI 抓取有特殊作用。",[11,554,555,558],{},[35,556,557],{},"表格",":对比类内容用表格,被引用率极高。AI 抓对比类问题(如\"XX 和 YY 的区别\")时,优先引用表格。",[11,560,561,249],{},[35,562,563],{},"Before(散文式对比)",[251,565,566],{},[11,567,568],{},"SEO 和 GEO 的区别主要体现在以下几个方面。首先是目标不同,SEO 是让搜索引擎排名靠前,GEO 是让 AI 引用你。其次是目标引擎不同,SEO 面向百度和 Google,GEO 面向 ChatGPT 和 Perplexity。再次是核心信号不同...",[11,570,571,249],{},[35,572,573],{},"After(表格对比)",[575,576,577,592],"table",{},[578,579,580],"thead",{},[581,582,583,587,590],"tr",{},[584,585,586],"th",{},"维度",[584,588,589],{},"SEO",[584,591,226],{},[593,594,595,607,618,629],"tbody",{},[581,596,597,601,604],{},[598,599,600],"td",{},"目标",[598,602,603],{},"搜索结果排名靠前",[598,605,606],{},"AI 回答时引用你",[581,608,609,612,615],{},[598,610,611],{},"目标引擎",[598,613,614],{},"百度、Google、Bing",[598,616,617],{},"ChatGPT、Perplexity、文心一言",[581,619,620,623,626],{},[598,621,622],{},"核心信号",[598,624,625],{},"关键词、外链、技术 SEO",[598,627,628],{},"内容结构化、权威度、可引用性",[581,630,631,634,637],{},[598,632,633],{},"见效周期",[598,635,636],{},"3-6 个月",[598,638,639],{},"2-4 个月",[11,641,642],{},"AI 抓表格的概率比散文高得多。我们监测过一个 GEO 案例,改成表格后,ChatGPT 引用率从 12% 涨到 34%。",[11,644,645,648],{},[35,646,647],{},"列表",":步骤类、条件类、并列要点用列表。AI 抓\"如何做 XX\"类问题时,优先抓列表。",[11,650,651,249],{},[35,652,653],{},"适合列表的内容",[144,655,656,659,662],{},[147,657,658],{},"步骤(1、2、3、4)",[147,660,661],{},"条件(满足 A、B、C 才能做)",[147,663,664],{},"要点(几个并列的点)",[11,666,667,249],{},[35,668,669],{},"不适合列表的内容",[144,671,672,675,678],{},[147,673,674],{},"论证(需要展开论述)",[147,676,677],{},"因果关系(需要解释为什么)",[147,679,680],{},"复杂对比(用表格更清晰)",[15,682,683],{"id":683},"一个综合例子",[11,685,686],{},"我们给一个客户改 GEO 内容,1 个月内 ChatGPT 引用率从 8% 涨到 31%。改的内容:",[688,689,690,693,696,699,702,705],"ol",{},[147,691,692],{},"开头 100 字改为结论前置(技巧 1)",[147,694,695],{},"关键段落加可引用事实句(技巧 2)",[147,697,698],{},"数据加出处和年份(技巧 3)",[147,700,701],{},"FAQ 答案压缩到 50 字以内(技巧 4)",[147,703,704],{},"段落拆短,首句改主题句(技巧 5)",[147,706,707],{},"对比内容改表格,步骤改列表(技巧 6)",[11,709,710],{},"工作量不大,大概 8 小时改 20 篇内容。效果立竿见影。",[15,712,713],{"id":713},"总结",[11,715,716],{},"GEO 的 6 个内容结构技巧:结论前置、可引用事实句、数据带出处、FAQ 答案简短、段落主题句、表格列表化。每个技巧都有 before\u002Fafter,改完就能看到引用率提升。GEO 不需要重写内容,结构改造是性价比最高的优化。",[11,718,719],{},"想知道你的内容在 AI 引擎里被引用率怎么样,联系我们做个免费诊断。",{"title":70,"searchDepth":71,"depth":71,"links":721},[722,723,724,725,726,727,728,729],{"id":239,"depth":71,"text":240},{"id":287,"depth":71,"text":288},{"id":356,"depth":71,"text":357},{"id":436,"depth":71,"text":437},{"id":497,"depth":71,"text":498},{"id":548,"depth":71,"text":549},{"id":683,"depth":71,"text":683},{"id":713,"depth":71,"text":713},"2026-01-05","GEO 的核心是让内容容易被 AI 引用。结论前置、可引用事实句、数据带出处、FAQ 结构、段落主题句、表格列表,6 个技巧带 before\u002Fafter 例子。",{},"\u002Fblog\u002Fgeo-content-structure",{"title":231,"description":731},"blog\u002Fgeo-content-structure",[226,737,738,739],"内容结构","AI引用","生成式引擎优化","WbrNI6JgVRoD7TiwYHlAeSCIrGc9pKDEExBXP05ll6U",1794088109491]