reddit · SaaS · 09-16 05:50 52 分 ← 返回列表

How I count assistant visits that arrive with no referrer

I count deep entries with no referrer as their own line in the report. Here's how. I take two lists of pages. The first list comes from the report and holds every page that got direct entries this month. The second list holds the pages that answer questions about our category. I keep the pages that sit on both lists. Nobody types a deep URL for a blog post from memory. So if that page also answers a category question, the visit most likely came from an assistant that didn't pass a referre

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：AI 助手/LLM 带来的访问常不带 referrer，在分析报告中显示为 direct 或 0，导致 SaaS 无法归因真实流量来源。

目标用户：SaaS 增长/营销/分析人员、独立开发者、内容站站长

证据：I count deep entries with no referrer as their own line in the report. His direct deep entries lasted seconds. The visits he could tie to an assistant lasted minutes. It's rough. It's still better than a zero sitting in the report where a real channel should be.

变现：提供 AI 流量归因分析插件或 SaaS，按站点/事件订阅收费；也可作为现有分析工具的付费模块。

MVP：连接 GA4/Plausible API，取无 referrer 的深层入口页面，与高意图分类页面取交集，按页面停留时长过滤机器人，输出 AI 流量估算报表；2 周可做脚本+简单 dashboard。

竞品：Google Analytics 4、Plausible、Fathom、Segment、Adobe Analytics

摘要：作者分享一种粗略方法：把无 referrer 的深层入口页与品类问答页取交集，并按页面停留时长过滤机器人，以估算 AI 助手带来的访问。一位创始人用自己的七个页面测试，发现直访深层页只停留几秒，而可归因 AI 的访问停留数分钟。该方法尚不严谨，但说明分析报告中 AI 流量被记为 0 是一个真实痛点。