reddit · SaaS · 09-22 06:50 57 分 ← 返回列表

I counted AI crawler hits on 4 small B2B sites for 27 days: 2,817 verified fetches, 5 referred visits

Everyone here is talking about getting cited by ChatGPT and Perplexity. I got curious about what that looks like from the server side, so I went through the logs of four small B2B sites I run, over 27 days. 2,817 verified fetches from the crawlers AI assistants use to read pages before citing them. 5 visits that arrived with an AI referrer. That's the first thing I didn't expect. Being read by the engine and getting a visitor from it are two separate events, and the gap between them is hu

五维评分

付费意愿

需求真实度

个人可落地

竞争空间

新鲜度

AI 分析

痛点：B2B 站点无法区分真实的 AI 爬虫抓取与伪造流量，且 AI 引用与带来访问之间差距巨大，导致 AI 搜索优化缺乏可靠指标。

目标用户：B2B SaaS 的内容营销、SEO/GEO 负责人和独立站长。

证据：85% of them failed. Scrapers love wearing an AI crawler's user agent. So if your "AI traffic" dashboard count user agents, most of what it shows you is fake, and the real number is a lot smaller and a lot more boring. Questions that named a competitor ("X vs Y", "alternatives to Y") triggered one 84% of the time.

变现：订阅制 SaaS：按站点或日志量收费的 AI 爬虫验证与 AI 搜索分析工具，附加对比页/替代页内容推荐。

MVP：2 周内做一个日志上传或接入工具，校验 AI 爬虫 IP 与 UA，区分真假抓取，统计 AI 引用与 referrer 访问，并基于竞品关键词生成对比/替代页选题建议。

竞品：Profound、Peec AI、Otterly.ai、Athena、Goodie AI

摘要：作者统计 4 个小型 B2B 站点 27 天日志，发现 2,817 次验证后的 AI 爬虫抓取仅带来 5 次 AI 来源访问；自称 AI 爬虫的请求中 85% 验证失败。ChatGPT 对品类问题仅约 10% 触发联网搜索，但对竞品对比/替代问题触发率高达 84%，说明真实 AI 流量监测和对比页优化存在工具机会。