# Everyone is talking about Cal AI's delisting. The real story is worse.
**作者**: Lucas Patiri
**日期**: 2026-04-16T16:32:24.000Z
**来源**: [https://x.com/lucaspatiri_/status/2044816187377242223](https://x.com/lucaspatiri_/status/2044816187377242223)
---
Yesterday Apple pulled Cal AI from the App Store. $50M ARR company, acquired by MyFitnessPal last month (btw for at least 200-250M I guess). Gone overnight because they used in-app payment sheets to bypass IAP.
The response Ive seen all day is split between two takes:
- "Apple is evil, this is anti-competitive"
- "Never bypass IAP, use Apple's 30%, end of story"
Both miss it
The truth is that the workaround Cal AI was using solved a problem apps actually have. That problem didnt go away when Apple pulled them. And there's a completely legal playbook that does everything payment sheets did, without the platform risk.
That playbook is web-to-app funnels. But the reason to run one isn't what most people think. The fee savings are smaller than you'd expect. Its the attribution, the retention flows, the cash flow. Things iOS never lets you touch.
Here's what web-to-app funnels actually are, why they work, and 12 tips to build one.
What a web-to-app funnel actually is.
A web-to-app funnel is a full acquisition and monetization flow that happens on the web, before the user ever opens your app.
The flow:
1. User sees an ad on TikTok, Meta, or YouTube
2. Ad clicks through to a web landing page, not the App Store
3. On the web the user goes through your onboarding (quiz, personalization, before-after, etc.)
4. At the end they hit a paywall, on the web
5. They pay with Stripe. Apple Pay still works. Card still works
6. After payment they get a deep link to the App Store and an email with login credentials
7. They download the app and open it as a paying user
Revenue never touches Apple. The 30% stays in your pocket. Stripe fees are around 3%.
Why Apple doesnt touch this.
The reason payment sheets got Cal AI killed is that the purchase happened inside the iOS app.
If your user pays on your website before downloading, that revenue is legally outside Apple's control. This is why web-to-app funnels are not a grey area. They are completely clear.
The part nobody wants to hear: the economics are closer than you think. If you're expecting a 27% ARPU lift just from switching to Stripe, you won't get it. Based on multiple studies, the conversion drop almost exactly cancels the fee savings. So why run a web funnel at all? Two reasons that have nothing to do with the 30%.
Attribution. iOS 14.5 broke mobile attribution in 2021 and it hasn't recovered. Web gives you the clean pixel data you need to actually scale paid media.
Retention control. Apple won't let you survey cancellers, offer retention discounts, or run save flows. Stripe lets you save up to 30% of churn.
Cash flow. Apple pays you monthly, with a 30 to 45 day holding window after each purchase. Stripe pays you daily, usually within 2 business days. For apps scaling on paid media this is key
Thats the real LTV lift, not the fee math. Thomas Petit, one of the sharpest app growth consultants out there, puts it clean: "People are going to web-to-app for the wrong reasons." The fees are not the game changer, the control of the customer relationship is.
> Now the 12 tips.
## Tip 1: The paywall belongs on the web, not in the app.
Most apps still put the paywall inside the app. That's a leftover habit from when all paid conversion had to happen inside iOS. It's no longer true.
If your best-performing paywall lives inside the app, clone it to the web and test it as the endpoint of a web-to-app funnel. You'll see conversion shifts immediately.
## Tip 2: Treat the web as the front of your funnel, the app as the product.
The shift is mental. The web is no longer a "marketing site". It's the acquisition AND monetization layer. The app is where the value gets delivered, not where payment happens.
Apps that internalize this shift build separate funnels for web and for App Store traffic. App Store traffic gets the IAP paywall. Web traffic gets the web paywall. Each is optimized for its own economics.
## Tip 3: Your onboarding quiz is the actual paywall.
Cal AI had 30 onboarding screens before a single payment prompt. That flow was not "onboarding", it was conversion infrastructure.
By the time they hit the paywall, they'd invested enough effort that paying felt like finishing the process, not starting a new one.
This works 10x better on the web because you can A/B test every screen on the fly. In the App Store, testing onboarding requires a full app update cycle.
## Tip 4: Collect the email before the paywall, not after.
On the web, ask for email at screen 2 or 3 of the quiz, not at the paywall. Reason: you want a retargeting list of everyone who started the quiz but didnt finish.
Abandoned-quiz email flows recover 10 to 20% of users who bounce. That recovery is impossible with in-app onboarding.
## Tip 5: Test paywall price by country.
On IAP you're locked into Apple's pricing tiers and can only adjust via App Store Connect. On the web you can geo-target prices dynamically and test incremental lift.
Typical wins:
- +15 to 30% price in English-speaking countries (US, UK, AU, CA)
- -30 to 50% price in LATAM and SEA
## Tip 7: The deep link after payment matters more than the paywall.
After payment you need a bulletproof flow to get the user into the app as an already-paid subscriber:
1. Redirect to a thank-you page with a "Get the app" button
2. Button deep-links to your App Store listing
3. Also email them a login magic link
4. First time they open the app, the app checks the web-authed session and unlocks premium
If this flow breaks you lose the customer after they paid. Every payment step must have a fallback (SMS, email, clipboard copy of a code, etc.).
## Tip 8: Attribution is cleaner on the web, use it.
On the web you have cookies, UTMs, pixel events, and full funnel visibility. You can see exactly which ad, creative, and campaign drove each paying subscriber.
This is the single biggest reason creators and performance teams push for web funnels. You can actually measure what's working.
## Tip 9: Run the comparison math explicitly.
For any campaign you run, calculate real CAC / LTV in both flows.
App Store traffic:
- CPI (cost per install)
- Install to subscriber conversion
- IAP revenue (minus Apple's 30%)
- LTV
Web funnel traffic:
- CPC (cost per click)
- Click to subscriber conversion
- Stripe revenue (minus 3%)
- LTV
## Tip 10: Dont migrate, duplicate.
Apps that try to "migrate" from IAP to web funnels usually break their App Store paying base in the process.
The correct move is to run both in parallel. IAP stays intact for App Store organic and ASO traffic. Web funnel becomes the endpoint for all paid media (TikTok, Meta, YouTube). The two co-exist.
Once the web funnel is proven, you can push harder on paid media because unit economics are better. Organic and ASO keep their own lifecycle.
## Tip 11: Your landing page needs to convert like a paywall, not a marketing site.
The biggest mistake in web-to-app funnels is treating the landing page like a product marketing site. "Features, testimonials, FAQ, pricing tiers".
It should look like your best-performing paywall, repurposed. Same hero message, same before-after, same social proof, same urgency. The landing page IS the top of the paywall. Treat it that way.
## Tip 12: Build the web version of your core product, even if it's lite.
This is the unlock nobody talks about. If your app has a web version of the core feature (even a stripped-down one), conversion jumps dramatically.
Why: the user can try the product before committing. They convert on the web, try it for 5 minutes, love it, then download the full app. You've reduced the trial-to-paid friction by eliminating the "install before I know if this is good" barrier.
## Tip 13: The $1M revenue threshold changes the math completely.
Apple's Small Business Program charges 15% instead of 30% for apps under $1M in annual revenue. If that's you, the math flips.
For apps under $1M, web funnels lose money on a per-user basis. Not break even. Lose money.
The web funnel play starts making sense when you cross $1M and Apple's cut jumps to 30%. Before that threshold, IAP plus a great paywall will beat a web funnel every time, even in net terms.
> The correct read for most early-stage apps: stay on IAP, obsess over paywall conversion, build web-to-app infrastructure in parallel, flip the switch when revenue crosses the threshold.
The bigger frame.
Cal AI didn't die because they were greedy or because Apple got mean. Cal AI got down because they picked the wrong vehicle for the right destination.
The destination was: keep more of the revenue from your paying users. That destination is correct. Every mobile founder should care about it.
This was always going to be shut down and it'll get worse the next months. It just took Apple 18 months.
The right vehicle has been available the whole time. Its web-to-app funnels. Not because you keep the 30%, that gain is smaller than the hype suggests. Because you keep the customer. Attribution fires cleanly. Retention flows are in your hands.
Stripe payment sheets were a shortcut to capture the same economics web funnels have given apps for years.
If your app is still running the entire monetization flow inside iOS in 2026, you're not protecting yourself from platform risk. You're volunteering to pay a 30% tax that your competitors have already opted out of. Legally.
Cal AI's mistake was not that they tried to escape the 30%. Their mistake was that they tried to escape it from inside iOS.
> The exit door has always been on the web.
## 相关链接
- [Lucas Patiri](https://x.com/lucaspatiri_)
- [@lucaspatiri_](https://x.com/lucaspatiri_)
- [120K](https://x.com/lucaspatiri_/status/2044816187377242223/analytics)
- [Upgrade to Premium](https://x.com/i/premium_sign_up)
- [12:32 AM · Apr 17, 2026](https://x.com/lucaspatiri_/status/2044816187377242223)
- [120.3K Views](https://x.com/lucaspatiri_/status/2044816187377242223/analytics)
- [View quotes](https://x.com/lucaspatiri_/status/2044816187377242223/quotes)
---
*导出时间: 2026/4/17 15:03:56*
---
## 中文翻译
# 人人都在谈论 Cal AI 被下架。真实情况更糟糕。
**作者**：Lucas Patiri
**日期**：2026-04-16T16:32:24.000Z
**来源**：[https://x.com/lucaspatiri_/status/2044816187377242223](https://x.com/lucaspatiri_/status/2044816187377242223)
---
昨天苹果将 Cal AI 从 App Store 下架。这是一家年经常性收入 5000 万美元的公司，上个月刚被 MyFitnessPal 收购（顺便说一句，我猜收购价至少 2-2.5 亿美元）。一夜之间消失，因为他们使用应用内支付页面绕过了 IAP。
我一整天看到的反应分为两种观点：
- “苹果是邪恶的，这是反竞争行为”
- “永远不要绕过 IAP，老老实实给苹果 30%，就这样”
这两种观点都没说到点子上
真相是，Cal AI 使用的变通方法解决了一个应用确实存在的问题。当苹果把它们下架时，这个问题并没有消失。而且有一套完全合法的方案，能实现支付页面所做的一切，且没有平台风险。
这套方案就是 Web 转应用漏斗。但运营它的原因并非大多数人所想。节省的费用比你预期的要少。真正重要的是归因、留存流程、现金流。这些是 iOS 从不允许你触碰的东西。
下面介绍 Web 转应用漏斗究竟是什么、为什么有效，以及搭建它的 12 个技巧。
Web 转应用漏斗究竟是什么。
Web 转应用漏斗是一套完整的获客和变现流程，全部发生在网页上，在用户打开你的应用之前。
流程如下：
1. 用户在 TikTok、Meta 或 YouTube 上看到广告
2. 广告点击后进入网页落地页，而不是 App Store
3. 在网页上，用户完成你的引导流程（测验、个性化设置、前后对比等）
4. 最后他们遇到付费墙，在网页上
5. 他们用 Stripe 支付。Apple Pay 仍然可用。信用卡仍然可用
6. 支付后，他们收到一个指向 App Store 的深度链接，以及一封包含登录凭证的邮件
7. 他们下载应用，以已付费用户的身份打开它
收入永远不会经过苹果。30% 的抽成留在你的口袋里。Stripe 的费率约为 3%。
为什么苹果不会动这套方案。
支付页面导致 Cal AI 被下架的原因是购买发生在 iOS 应用内部。
如果你的用户在下载之前就在你的网站上付费，那么这部分收入在法律上不在苹果的管控范围内。这就是为什么 Web 转应用漏斗不是灰色地带。它完全合规。
没人愿意听的部分：经济账比你想的更接近。如果你指望仅靠切换到 Stripe 就获得 27% 的 ARPU 提升，你不会得到。多项研究表明，转化率的下降几乎完全抵消了费用节省。那为什么还要运营 Web 漏斗？有两个与 30% 无关的理由。
归因。iOS 14.5 在 2021 年破坏了移动归因，至今没有恢复。Web 能给你提供干净的像素数据，让你真正扩展付费媒体。
留存控制。苹果不会让你调查取消订阅的用户、提供留存折扣或运行挽回流程。Stripe 能让你挽回多达 30% 的流失用户。
现金流。苹果按月向你付款，每次购买后有 30 到 45 天的留存窗口。Stripe 每天向你付款，通常在 2 个工作日内到账。对于靠付费媒体扩张的应用来说，这是关键
这才是真正的 LTV 提升，而不是费用账。Thomas Petit——最敏锐的应用增长顾问之一——说得很直白：“人们因为错误的原因转向 Web 转应用。”费用不是改变游戏规则的因素，对客户关系的掌控才是。
> 现在是 12 个技巧。
## 技巧 1：付费墙应该放在网页上，而不是应用内。
大多数应用仍然把付费墙放在应用内。这是过去所有付费转化都必须在 iOS 内完成时留下的习惯。现在情况已经不同了。
如果你表现最好的付费墙在应用内，把它复制到网页上，并将其作为 Web 转应用漏斗的终点进行测试。你会立刻看到转化率的变化。
## 技巧 2：把网页当作漏斗的前端，把应用当作产品。
这是一种思维转变。网页不再是一个“营销网站”。它是获客和变现层。应用是交付价值的地方，而不是发生支付的地方。
内化这种转变的应用会为网页流量和 App Store 流量分别建立漏斗。App Store 流量使用 IAP 付费墙。网页流量使用网页付费墙。各自针对自己的经济模型进行优化。
## 技巧 3：你的引导测验才是真正的付费墙。
Cal AI 在第一次付款提示之前有 30 个引导页面。那个流程不是“引导”，而是转化基础设施。
当他们到达付费墙时，他们已经投入了足够的精力，以至于付费感觉像是完成流程，而不是开始新的流程。
这在网页上效果好 10 倍，因为你可以实时对每个页面进行 A/B 测试。在 App Store 中，测试引导流程需要一个完整的应用更新周期。
## 技巧 4：在付费墙之前收集邮箱，而不是之后。
在网页上，在测验的第 2 或第 3 页请求邮箱，而不是在付费墙。原因：你需要一个再营销列表，包含所有开始了测验但没有完成的人。
放弃测验的邮件挽回流程能挽回 10% 到 20% 的流失用户。这种挽回在应用内引导流程中是不可能实现的。
## 技巧 5：按国家测试付费墙价格。
在 IAP 中，你被锁定在苹果的定价层级，只能通过 App Store Connect 调整。在网页上，你可以动态地按地区设置价格并测试增量提升。
典型的优化：
- 在英语国家（美国、英国、澳大利亚、加拿大）提价 15% 到 30%
- 在拉美和东南亚降价 30% 到 50%
## 技巧 7：支付后的深度链接比付费墙更重要。
支付后，你需要一个万无一失的流程让用户以已付费订阅者的身份进入应用：
1. 重定向到带有“获取应用”按钮的感谢页面
2. 按钮深度链接到你的 App Store 页面
3. 同时通过电子邮件发送登录魔法链接
4. 用户首次打开应用时，应用会检查网页端已验证的会话并解锁高级功能
如果这个流程中断，你就会在用户付完款后失去他们。每一个支付步骤都必须有备用方案（短信、邮件、复制验证码到剪贴板等）。
## 技巧 8：Web 端的归因更干净，要好好利用。
在 Web 端，你有 Cookie、UTM 参数、像素事件以及完整的漏斗可见性。你可以清楚地看到是哪条广告、哪个创意、哪个活动带来了每一个付费订阅用户。
这是创作者和效果营销团队力推 Web 漏斗的最大原因。你能够真正衡量什么在起作用。
## 技巧 9：明确地做对比计算。
对于你投放的任何一个活动，都要计算两条路径下的真实 CAC / LTV。
App Store 流量：
- CPI（单次安装成本）
- 安装到订阅的转化率
- IAP 收入（减去 Apple 的 30%）
- LTV
Web 漏斗流量：
- CPC（单次点击成本）
- 点击到订阅的转化率
- Stripe 收入（减去 3%）
- LTV
## 技巧 10：不要迁移，要复制。
那些试图从 IAP “迁移”到 Web 漏斗的 App，通常会在此过程中搞坏他们在 App Store 的付费用户基础。
正确的做法是两者并行运行。IAP 保持原样，服务于 App Store 的自然流量和 ASO 流量。Web 漏斗则成为所有付费媒体（TikTok、Meta、YouTube）的终点。两者共存。
一旦 Web 漏斗被验证，你就可以在付费媒体上加码，因为单位经济模型更好。自然流量和 ASO 保持它们自己的生命周期。
## 技巧 11：你的落地页需要像付费墙一样能转化，而不是像营销官网。
Web 转 App 漏斗中最大的错误，就是把落地页当成产品营销网站来做。“功能介绍、用户评价、常见问题、定价档位”。
它应该看起来像你表现最好的付费墙的改版。同样的主标语、同样的前后对比、同样的社会证明、同样的紧迫感。落地页就是付费墙的顶端。要这样对待它。
## 技巧 12：构建核心产品的 Web 版本，即使是精简版。
这是没人谈论的关键。如果你的 App 的核心功能有 Web 版本（哪怕是个精简版），转化率会大幅跃升。
原因：用户可以在做出承诺之前先试用产品。他们在 Web 端完成转化，试用 5 分钟，喜欢上它，然后再下载完整的 App。你通过消除“在不知道好不好用之前就得先安装”的障碍，降低了试用到付费的摩擦。
## 技巧 13：100 万美元的收入门槛会彻底改变这笔账。
Apple 的小型企业计划对年收入低于 100 万美元的 App 只收取 15% 而不是 30% 的佣金。如果你属于这种情况，那笔账就反过来了。
对于年收入低于 100 万美元的 App，Web 漏斗在单用户层面是亏钱的。不是持平，是亏钱。
Web 漏斗这个玩法是在你跨过 100 万美元、Apple 的抽成跳到 30% 之后才开始有意义的。在那道门槛之前，IAP 加上一个优秀的付费墙，每次都能胜过 Web 漏斗，即使算净利润也是如此。
> 对于大多数早期 App，正确的解读是：留在 IAP，死磕付费墙转化率，并行构建 Web 转 App 的基础设施，等收入跨过门槛再切换。
更大的格局。
Cal AI 不是因为贪婪而倒下，也不是因为 Apple 变得刻薄。Cal AI 出局是因为他们为正确的目的地选错了交通工具。
这个目的地是：从付费用户身上留住更多的收入。这个目的地是正确的。每一位移动端创始人都应该关心它。
这种做法本来就注定会被封杀，而且接下来几个月情况会更糟。Apple 只是用了 18 个月。
正确的交通工具一直都在，那就是 Web 转 App 漏斗。不是因为你能留住那 30%，这个收益比炒作所暗示的要小。而是因为你能留住客户。归因可以精准触发。留存流程掌握在你自己手中。
Stripe 支付页只是个捷径，试图截取 Web 漏斗多年来一直提供给 App 的同样经济收益。
如果你的 App 在 2026 年仍然把整个变现流程都放在 iOS 内运行，那你不是在保护自己免受平台风险，你是在自愿缴纳一笔你的竞争对手早已合法免除的 30% 税。
Cal AI 的错误不在于他们试图逃离那 30%。他们的错误在于，他们试图从 iOS 内部逃离。
> 出口的大门一直都在 Web 上。
## 相关链接
- [Lucas Patiri](https://x.com/lucaspatiri_)
- [@lucaspatiri_](https://x.com/lucaspatiri_)
- [120K](https://x.com/lucaspatiri_/status/2044816187377242223/analytics)
- [升级到 Premium](https://x.com/i/premium_sign_up)
- [12:32 AM · Apr 17, 2026](https://x.com/lucaspatiri_/status/2044816187377242223)
- [120.3K Views](https://x.com/lucaspatiri_/status/2044816187377242223/analytics)
- [View quotes](https://x.com/lucaspatiri_/status/2044816187377242223/quotes)
---
*导出时间: 2026/4/17 15:03:56*
