# How to find & fix what's killing your sales (with the Claude Buyer Simulation skill)
**作者**: Ole Lehmann
**日期**: 2026-04-24T13:26:21.000Z
**来源**: [https://x.com/itsolelehmann/status/2047668468904808531](https://x.com/itsolelehmann/status/2047668468904808531)
---
Whatever you use to sell, your landing pages, your emails, or your ads, it all makes perfect sense to you. Because you made it.
But your audience sees something completely different.
So I built a skill that simulates 5 different buyer personas reading whatever you give it, section by section.
It tells you exactly who would buy, who wouldn't, and what to fix so more people convert.
In this article I'll show you exactly how to set it up.
Here's the method and the exact skill I built, so you can run it yourself:
P.S. If you want more AI workflows like this one delivered to your inbox every week, join 37k readers getting them free:
aisolo.beehiiv.com/subscribe
## You'll never see what your buyer sees
Of course it all makes sense to you. You know the product inside out:
- the page
- the emails
- the backstory
- the pricing logic
- why it's worth every dollar
But your buyer has none of that context.
They came to your page after a long day of getting hammered by their boss, fresh off 45 minutes of doomscrolling Instagram.
They're skimming with half their attention. So the second something doesn't add up, they're gone.
All you get back is a conversion rate: a number that says something is wrong but never says what.
And every day that page is live with those blind spots, you're:
- Spending ad dollars sending people to something that's leaking
- Burning audience trust with an offer that confuses them
- Losing sales you'll never know you lost
…all because of a vague headline or a missing proof point you couldn't see from the inside.
You need simulated buyers who can test it from the outside.
## Where this comes from
Shopify had this exact problem at scale.
A merchant changes their storefront. How do they know if it converts better?
The old way: A/B test it, but then you have to wait weeks. And it’s ineffective for small merchants because they don't have enough traffic for testing to work at all.
So Shopify built SimGym. They send hundreds of simulated shoppers to browse a merchant's store. Each shopper has a persona, a budget, and a shopping intent.
Each runs in a real cloud browser, sees the page, decides what to do, clicks, scrolls, adds to cart, or leaves. Basically: a simulated A/B test in minutes instead of weeks.
I rebuilt the concept as a Claude skill. Same idea: simulate buyers walking through whatever you're putting in front of people and reporting back what they'd actually do.
Their version runs on dedicated AI infrastructure that costs millions.
This one is free, gets you the 80/20 result, and runs inside Claude in about 3 minutes.
I call it "buy or bounce."
## The 5 buyers
Each buyer represents a different reason people don't convert.
The Shut-up-and-take-my-Money Buyer Has the problem, has the money, and is actively shopping. If they bounce, something fundamental is broken. This Buyer is your canary. If they won't convert, nobody will.
The Skeptic Been burned before. Every bold claim triggers their BS detector. Scanning for proof: specific numbers, case studies, guarantees, anything that separates real from hype.
The Price-Conscious Buyer Wants the outcome but doing the math on whether it's worth the money the entire time. They need the value to clearly and obviously exceed the cost.
The Confused Visitor Found your page but genuinely can't tell what this is or if it's for them. They'd buy if they understood the offer (and this is the most common and most fixable problem).
The Comparison Shopper Evaluating 2-3 alternatives. Reading your page with one question: "Why this and not that?"
These 5 map to the core reasons people don't buy: broken fundamentals, missing credibility, weak value justification, unclear positioning, no differentiation.
## Why section-by-section matters
This is what separates buy or bounce from just asking Claude to review your page.
Each buyer reads what you give it top to bottom and narrates their internal monologue at every section:
- "Okay, this headline grabs me, I have this problem..."
- "Wait, what does 'transform your workflow' actually mean? That's vague..."
- "These features sound nice but how is this different from the other 3 things I'm looking at..."
- "That price is higher than I expected for something I still don't fully understand..."
- "I'm not clicking this."
You see the exact moment each buyer's interest dies.
And because all 5 walk through the same material, it maps where the friction clusters. If 3 out of 5 bounced at the same section, that's your biggest leak.
Then it tells you exactly what to fix, in what order, with specific rewrite suggestions.
## How to set it up
Grab the skill here. Drop it into your Skills in Claude Code or Cowork (Customize > Skills > Add skill).
Say "buy or bounce this" and paste whatever you want tested. Landing page URL, email, sales page, ad creative, whatever you're trying to get someone to say yes to.
The skill parses it into sections, spawns all 5 buyers in parallel, and each one walks through narrating their buying decision.
You get a visual HTML report: conversion score, section-by-section friction map, prioritized fixes with specific rewrites, and full buyer walkthroughs if you want to dig in.
## What happened when I tested my own landing page
I ran buy or bounce on one of my workshop landing pages (now closed, but you can view the page here: https://aisolopreneur.lpages.co/claude-cowork-workshop/)
3 out of 5 would buy. 2 on the fence. Nobody bounced.
The Shut-up-and-take-my-Money Buyer converted. The "click-by-click" promise in the subheadline hooked them, and the testimonials sealed it. Angelo going from 5 figures on web designers to doing it himself in hours, Tanya taking a landing page from concept to live in under 2 hours.
The Price-Conscious Buyer converted. They did the math: if $197 saves even one project that would've cost $2-5k in freelancer fees, the ROI is obvious. Their words: "If even half of those results are real for me, $197 pays for itself on the first project."
The Comparison Shopper converted. They'd seen other AI courses and YouTube tutorials but none showed specific, measurable outcomes from real students.
The Skeptic was on the fence. The testimonials and proof story were credible enough to get past initial skepticism, but they wanted to actually see video proof of the 8-minute landing page creation before committing $197. If the story holds up on camera, they're buying.
The Confused Visitor was on the fence. They understood what the workshop teaches you to do, but never fully grasped what Claude Cowork itself is. Is it software? A framework? A way of prompting?
Then the synthesis mapped the patterns and gave me specific advice:
- Add one plain-language sentence early on the page explaining what Claude Cowork actually is. The Confused Visitor couldn't tell if it's software, a framework, or a method.
- Add a video demo or screenshots showing the 8-minute landing page creation. The Skeptic needed to see it, not just read about it.
- Double down on testimonials with specific cost/time savings. Angelo's 5 figures saved and Tanya's 2-hour turnaround did the heavy lifting for 2 buyers.
- Clarify the "10M+ people" claim in the subheadline. Multiple buyers couldn't tell if it's my reach or a promise about what they'd achieve.
- Replace the "falling behind" stakes section with aspiration-based framing. 2 buyers came in curious, not panicked, and the fear angle pushed them away.
A prioritized list of what to fix and in what order. All in about 3 minutes.
P.S. If you want to implement AI workflows like this into your business, get them 100% free in your inbox here: aisolo.beehiiv.com/subscribe
## What else to test
Buy or bounce works on anything where you're trying to get someone to say yes.
- Landing pages before you drive traffic. Find what's broken before you spend money on a page that leaks.
- Email sequences before you hit send.
- Ad creatives before you spend budget. Ads have about 2 seconds to make sense to someone who's never heard of you.
- Sales pages before launch.
- Proposals and pitch decks before the meeting.
## Go test something
Think about the page, email, or ad you're about to publish. The one you've been going back and forth on.
Buy or bounce it. Grab the skill here. Say "buy or bounce this" and paste what you want tested.
3 minutes from now you'll know exactly where your buyers would drop off, what's confusing them, and what to fix first so more people actually buy.
P.S. If you want more AI workflows that help you get more customers, more attention, and more done (without working more hours)...
I send them to 37k readers every week for free. Plus you get a free Claude Cowork masterclass when you join: aisolo.beehiiv.com/subscribe
## 相关链接
- [Ole Lehmann](https://x.com/itsolelehmann)
- [@itsolelehmann](https://x.com/itsolelehmann)
- [2.2K](https://x.com/itsolelehmann/status/2047668468904808531/analytics)
- [aisolo.beehiiv.com/subscribe](https://aisolo.beehiiv.com/subscribe)
- [here](https://github.com/olelehmann1337/claude-skills/blob/main/skills/buy-or-bounce/SKILL.md)
- [https://aisolopreneur.lpages.co/claude-cowork-workshop/](https://aisolopreneur.lpages.co/claude-cowork-workshop/)
- [aisolo.beehiiv.com/subscribe](https://aisolo.beehiiv.com/subscribe)
- [aisolo.beehiiv.com/subscribe](https://aisolo.beehiiv.com/subscribe)
- [Upgrade to Premium](https://x.com/i/premium_sign_up)
- [9:26 PM · Apr 24, 2026](https://x.com/itsolelehmann/status/2047668468904808531)
- [2,257 Views](https://x.com/itsolelehmann/status/2047668468904808531/analytics)
- [View quotes](https://x.com/itsolelehmann/status/2047668468904808531/quotes)
---
*导出时间: 2026/4/24 22:32:37*
---
## 中文翻译
# 如何发现并修复扼杀你销售额的罪魁祸首（借助 Claude 买家模拟技能）
**作者**: Ole Lehmann
**日期**: 2026-04-24T13:26:21.000Z
**来源**: [https://x.com/itsolelehmann/status/2047668468904808531](https://x.com/itsolelehmann/status/2047668468904808531)
---
无论你使用什么方式进行销售——落地页、电子邮件还是广告——它们对你来说都完全合乎逻辑。毕竟是你制作的。
但你的受众看到的却是完全不同的东西。
所以我构建了一个技能，它可以模拟 5 种不同的买家画像，逐部分阅读你提供的任何内容。
它会准确告诉你谁会买，谁不会买，以及该修复什么才能让更多人转化。
在本文中，我将向你展示如何准确地进行设置。
以下是该方法以及我构建的确切技能，你可以亲自运行它：
P.S. 如果你想每周收到更多这样的 AI 工作流，欢迎加入 37,000 名读者的免费行列：
aisolo.beehiiv.com/subscribe
## 你永远看不到买家看到的东西
当然，这一切对你来说都很有道理。你对产品了如指掌：
- 页面
- 邮件
- 背景故事
- 定价逻辑
- 为什么它物有所值
但你的买家完全没有这些背景信息。
他们来到你的页面时，刚被老板折磨了一整天，或者是刚刷了 45 分钟 Instagram 的“末日刷屏”。
他们只投入了一半的注意力在浏览。所以，一旦有一点不合逻辑，他们就会离开。
你得到的只是一个转化率：一个告诉你有问题但从不告诉你具体问题是什么的数字。
而只要这个页面带着这些盲点在线一天，你就在：
- 浪费广告费，把人导向一个正在漏水的页面
- 用令人困惑的报价消耗受众的信任
- 丢失你永远无法知晓的销售机会
……这一切都只因为一个模糊的标题，或者一个因为你身在局中而看不到的缺失的证明点。
你需要能够从外部进行测试的模拟买家。
## 这个方法的由来
Shopify 在大规模运营时就面临过这个确切的问题。
商家修改了店面。他们怎么知道转化率是否提高了？
旧方法：A/B 测试，但那样你得等上好几周。这对于小商家来说效率很低，因为他们没有足够的流量让测试产生效果。
所以 Shopify 构建了 SimGym。他们向商家的商店发送数百个模拟购物者。每个购物者都有一个人设、预算和购物意图。
每个都在真实的云浏览器中运行，看到页面，决定做什么，点击、滚动、加购或离开。基本上就是：几分钟内完成模拟 A/B 测试，而不是几周。
我将其概念重建为一个 Claude 技能。同样的想法：模拟买家浏览你展示给他们的任何内容，并报告他们实际会做什么。
他们的版本运行在耗资数百万美元的专用 AI 基础设施上。
这个版本是免费的，能帮你获得 80/20 的结果，并且在 Claude 内大约 3 分钟就能运行完。
我称之为“买还是跳”。
## 5 种买家
每种买家代表人们不转化的不同原因。
**“闭嘴拿走我的钱”型买家**
有问题，有钱，并且正在积极购物。如果他们跳走了，说明基础东西坏了。这个买家是你的“金丝雀”（风向标）。如果他们都不转化，那没人会转化。
**怀疑论者**
以前被坑过。每一个大胆的声明都会触发他们的废话探测器。他们在扫描证据：具体数字、案例研究、保证，任何能区分真实与炒作的东西。
**价格敏感型买家**
想要这个结果，但整个过程都在算这笔钱是否值。他们需要价值明显且显而易见地超过成本。
**困惑的访客**
找到了你的页面，但真的看不出这是什么，或者是否适合他们。如果他们理解了这个报价，他们是会买的（这是最常见也最容易修复的问题）。
**比价者**
正在评估 2-3 个替代方案。阅读你的页面时脑子里只有一个问题：“为什么选这个不选那个？”
这 5 种对应了人们不购买的核心原因：基础损坏、缺乏可信度、价值证明薄弱、定位不清、没有差异化。
## 为什么逐部分很重要
这就是“买还是跳”与仅仅让 Claude 审查你的页面的区别所在。
每个买家都会从头到尾阅读你给的内容，并在每个部分叙述他们的内心独白：
- “好吧，这个标题吸引了我，我有这个问题……”
- “等等，‘改变你的工作流程’到底什么意思？太模糊了……”
- “这些功能听起来不错，但这跟我正在看的其他 3 个东西有什么区别……”
- “鉴于我还没完全理解这东西，这个价格比我预期的要高……”
- “我不会点这个。”
你能准确看到每个买家的兴趣在哪一刻熄灭。
而且因为所有 5 个买家都浏览了同样的材料，它能映射出摩擦点集中在哪。如果 5 个中有 3 个在同一部分跳走了，那就是你最大的漏洞。
然后它会确切地告诉你该修复什么，按什么顺序，并提供具体的重写建议。
## 如何设置
在这里获取该技能。将其放入你的 Claude Code 或 Cowork 的技能中（Customize > Skills > Add skill）。
说“buy or bounce this”（买还是跳这个），然后粘贴任何你想测试的内容。落地页 URL、邮件、销售页、广告创意，任何你试图让人点头同意的东西。
该技能会将其解析为几个部分，并行生成所有 5 个买家，每个都会边走边叙述他们的购买决策。
你会得到一份可视化的 HTML 报告：转化分数、逐部分的摩擦地图、带有具体重写建议的优先修复列表，如果你想深究，还有完整的买家浏览记录。
## 当我测试自己的落地页时发生了什么
我在我的一个工作坊落地页上运行了“买还是跳”（现已关闭，但你可以在这里查看页面：https://aisolopreneur.lpages.co/claude-cowork-workshop/）
5 人中有 3 人会买。2 人犹豫不决。没人跳走。
**“闭嘴拿走我的钱”型买家转化了。**
副标题中“点击式操作”的承诺吸引了他们，推荐信锁定了交易。Angelo 从花 5 位数请网页设计师到自己几小时内搞定，Tanya 将落地页从概念到上线不到 2 小时。
**价格敏感型买家转化了。**
他们算了一笔账：如果 197 美元能挽救哪怕一个本需支付 2-5 千美元自由职业费用的项目，投资回报率显而易见。他们的话：“如果哪怕只有一半的结果对我来说是真的，197 美元在第一个项目上就回本了。”
**比价者转化了。**
他们看过其他 AI 课程和 YouTube 教程，但都没有展示真实学生具体、可衡量的成果。
**怀疑论者在犹豫。**
推荐信和证明故事足以通过最初的怀疑，但他们希望在投入 197 美元之前，真正看到 8 分钟创建落地页的视频证明。如果故事在镜头前站得住脚，他们就会买。
**困惑的访客在犹豫。**
他们理解了工作坊教你做什么，但从未完全掌握 Claude Cowork 本身到底是什么。是软件？框架？还是一种提示词方法？
然后综合分析映射了模式，并给了我具体建议：
- 在页面早期添加一句通俗语言解释 Claude Cowork 到底是什么。困惑的访客分不清它是软件、框架还是方法。
- 添加视频演示或截图，展示 8 分钟创建落地页的过程。怀疑论者需要看到它，而不仅仅是读到它。
- 加大带有具体成本/时间节省的推荐信的比重。Angelo 省下的 5 位数和 Tanya 的 2 小时周转为 2 个买家完成了主要说服工作。
- 阐明副标题中“1000 万+人”的说法。多个买家分不清这是我的覆盖量还是对他们会达成目标的承诺。
- 用基于愿望的框架替换“落后于人”的风险部分。2 个买家是出于好奇进来的，而不是恐慌，恐惧角度反而把他们推开了。
一份包含修复内容和顺序的优先列表。全程大约 3 分钟。
P.S. 如果你想将这样的 AI 工作流实施到你的业务中，可以在这里免费 100% 获取：aisolo.beehiiv.com/subscribe
## 还能测试什么
“买还是跳”适用于任何你试图让人点头同意的场景。
- 在引流之前测试落地页。在花钱给一个漏水的页面之前，先找出哪里坏了。
- 在点击发送之前测试邮件序列。
- 在花费预算之前测试广告创意。广告只有大约 2 秒钟的时间让一个从未听说过你的人弄明白。
- 在发布之前测试销售页。
- 在会议之前测试提案和演示文稿。
## 去测试点什么
想想你正要发布的页面、邮件或广告。那个让你来回纠结的。
对它运行“买还是跳”。在这里获取技能。说“buy or bounce this”并粘贴你想测试的内容。
3 分钟后，你就会确切知道你的买家会在哪里流失，什么在困扰他们，以及首先修复什么才能让更多人真正购买。
P.S. 如果你想获得更多能帮你赢得更多客户、获得更多关注、完成更多工作（而无需加班）的 AI 工作流……
我每周都会免费发送给 37,000 名读者。此外，加入时你还能获得免费的 Claude Cowork 大师课：aisolo.beehiiv.com/subscribe