Readiness doesn't end at the border. It ends at checkout.准备度不会止步于国境线，它止步于收银台。

The compliance file didn’t reset. It had to be re-expressed.合规资料没有清零重来，只是需要换一种方式讲出来

Nothing in the underlying regulatory picture changes because the sale moved from a distributor's purchase order to a consumer's checkout. FALCPA — the federal Food Allergen Labeling and Consumer Protection Act — requires any retail food product to declare its major allergens, no matter which channel sells it.销售场景从经销商的采购订单变成消费者的收银台，底层监管要求并不会因此改变。FALCPA（美国联邦《食品过敏原标示与消费者保护法》）要求申报牛奶、坚果等主要致敏成分，这项过敏原声明要求适用于任何零售食品，无论渠道。

California's Prop 65 — a state law requiring a warning label on products sold there if they contain certain chemicals, including lead, above a set threshold — has its own rule for online sales: current as of the amended regulation effective January 1, 2025, the warning has to display before the sale completes. That can happen on the product page itself, through a clearly labeled link, or via a pop-up, but it has to name at least one specific chemical (for tea, almost always lead) in the warning itself. None of that is new law invented for this brand. It is the same obligation the factory already carried, now needing to live somewhere it never had to live before: a public product page for each individual SKU — short for stock keeping unit, meaning each distinct product listing — that a stranger reads before paying.加州 Prop 65——一项州法律，规定产品若含有铅等特定化学物质超过设定的量，在加州销售时就必须附警示标签——针对线上销售还有自己的一条规则：2025年1月1日起生效的修订版要求，警示必须在交易完成前展示。这个警示可以出现在产品页面上、通过清晰标注的链接、或通过弹窗，但必须点名至少一种具体化学物质（对茶而言，几乎总是"铅"）。这不是为这个品牌新发明出来的法规，而是工厂本就承担的同一份义务，如今需要出现在一个它从未需要出现的地方：一个陌生人付款前会读到的、公开的产品页面——而且是按每一个 SKU（库存单位，即每一个独立的商品条目）单独区分的产品页面。

Nutrition Facts status is the sharpest edge of this, because it is not a blanket answer. Plain loose-leaf tea can qualify for the FDA's exemption from a full Nutrition Facts panel; a matcha latte kit with added milk solids or a flavored blend with a calorie-contributing ingredient generally cannot. That means the storefront had to carry a different, correct answer on a SKU-by-SKU basis, not one compliance blurb copied across every product page. Getting that boundary right is a labeling determination, not a copywriting one.营养成分表的豁免状态是其中最需要精细处理的一点，因为它不是一个统一答案。纯散叶茶可以适用 FDA 对完整营养成分表的豁免；而添加了奶粉的抹茶拿铁套装，或含有贡献热量成分的调味茶，通常不能适用同一豁免。这意味着独立站必须按 SKU 逐一给出正确答案，而不是把同一段合规文案复制粘贴到每一个产品页面上。判断这条界限在哪里，是一项标签判定工作，不是文案工作。

So the build treated the compliance file as the foundation, not an afterthought bolted on before launch. Each product page was built to carry the Nutrition Facts status, the FALCPA allergen statement, and the Prop 65 short-form warning that specific SKU actually required, and the content was checked by the same category of labeling specialist who had signed off on the brand's original readiness dossier. Easterlies coordinated that review. It is not a certification body, not the brand's labeling authority, and did not make the underlying regulatory determinations itself.因此这次搭建把合规资料当作地基，而不是上线前临时补挂上去的附件。每一个产品页面都按该 SKU 实际需要的内容，标注营养成分表豁免状态、FALCPA 过敏原声明和 Prop 65 短式警示，内容由曾为该品牌最初的准备度资料包签字确认的同类标签专家复核。Easterlies 负责对接这次复核，但它不是认证机构，不是品牌的标签权威，也没有自己做出底层的监管判定。

A tin of tea is a low-frequency purchase. Retention has to do the work acquisition can’t.一罐茶，凭什么让人买第二次？

Premium loose-leaf and matcha are not impulse categories bought weekly. The average order value, or AOV, sits in a favorable band for direct-to-consumer conversion, but the natural repurchase cycle is slow, which means a brand that only sells one-off tins is rebuilding its customer base from zero every month. The lever that changes that arithmetic is the same one several real, operating tea brands have leaned on publicly: subscription. Recharge's own account of Tenzo Tea's shift from a wholesale-first business to a subscription-first DTC model describes exactly this logic, using a unified view of the customer across the storefront, subscription platform, and messaging channels to keep a buyer active instead of one-and-done.优质散叶茶和抹茶不是每周都会冲动购买的品类。它们的客单价（AOV，顾客平均每笔订单花费的金额）对独立站转化来说处在有利区间，但自然复购周期很慢，这意味着如果品牌只卖单次的单罐产品，它每个月都要从零重建自己的客户基础。能改变这道算术题的杠杆，正是好几个真实运营中的茶品牌公开采用过的同一个办法：订阅。Recharge 自己披露的 Tenzo Tea 案例——从以批发为主转向以订阅为主的独立站模式——描述的正是这套逻辑：把独立站、订阅平台和消息渠道里的客户信息打通成一个统一视图，让一次性买家变成持续活跃的订阅用户。

The build here followed the same shape: subscription mechanics configured for a reorder-every-N-weeks cadence on the core SKUs, so the brand's revenue stops depending entirely on new-visitor traffic. Underneath that sits the messaging layer that actually keeps people subscribed or brings them back when they lapse: automated email and SMS flows, meaning pre-built message sequences triggered by a shopper's own behavior, covering welcome, abandoned checkout, post-purchase replenishment, and win-back. Industry benchmark data on flow performance is consistent on one point: Klaviyo's own benchmark data puts flow-driven email revenue at roughly 41% of total email revenue from just 5.3% of sends, the kind of disproportionate return automated flows also tend to show on the SMS side relative to one-off campaign sends. That is the argument for building the flows correctly once rather than sending more campaigns later.这次搭建走的是同一条路：为核心 SKU 配置"每隔 N 周复购一次"节奏的订阅机制，让品牌的营收不再完全依赖新访客流量。订阅机制之下，是真正能留住订阅用户、或在用户流失后把人找回来的消息层——邮件与短信自动化流程，也就是根据顾客自己的行为（比如欢迎新用户、弃单、购后补货、召回流失用户）自动触发发送的一套预设消息，不需要人工一条条手动发。行业基准数据在一点上高度一致：Klaviyo 自己的基准数据显示，仅占发送量 5.3% 的自动化流程邮件，贡献了约 41% 的邮件总营收——这种不成比例的回报，短信侧的自动化流程相对于一次性群发campaign，通常也会体现出类似的趋势。这正是"把自动化流程一次性搭对"比"以后多发几次campaign"更值得投入的原因。

None of this was framed as a guaranteed retention number. Subscription attach rates and repeat-purchase benchmarks vary by category and by brand execution, and this build did not promise the brand would land at the top of that range. What it delivered was the mechanism, correctly wired, before a dollar of paid traffic arrived to test it.这一切都没有被包装成某个保证达到的留存数字。订阅转化率和复购率基准会因品类和品牌执行方式而异，这次搭建也没有承诺品牌一定能落在区间的顶端。它交付的是这套机制本身，在第一美元付费流量进来接受检验之前，先把线接对。

Paid social got a budget only after test purchases proved the tracking was real付费社媒只有在测试订单验证过追踪真实有效之后，才拿到预算

A storefront and a subscription flow are worthless if the traffic arriving at them is invisible to the brand's own reporting. Before any ad spend moved, the Meta and, where relevant, TikTok ad accounts were structured with conversion tracking, and that tracking was verified against real test purchases, actual orders placed and confirmed to attribute correctly, not a pixel installed and assumed to work. This is a deliberately unglamorous step, and it is the one most often skipped by brands eager to get spend live. Skipping it means the first weeks of ad spend buy data nobody can trust.如果独立站和订阅流程接收到的流量，在品牌自己的报表里根本看不见，那这一切都是空的。在任何广告费花出去之前，Meta（以及在相关的情况下 TikTok）广告账户就已经搭好了转化追踪，而且这套追踪是用真实测试订单——真正下单并确认归因正确的订单，而不是装上像素就假设它能用——验证过的。这是一个刻意显得不起眼的步骤，也是那些急着让预算跑起来的品牌最常跳过的一步。跳过它，意味着广告投放头几周买到的数据，没人能信。

The spend plan itself was sized to a 90-day validation window rather than treated as an open-ended budget, with a single reporting stack, sessions, conversion rate, AOV, blended customer acquisition cost by channel, subscription attach rate, and contribution margin, built so the day-30, day-60, and day-90 checkpoints could be read off a dashboard instead of argued about in a meeting. Contribution margin — what's left from a sale after subtracting the costs that scale with it, before subtracting fixed overhead — not gross margin, was the number that mattered, because early contribution margin is commonly thin or negative while acquisition cost is still being optimized and before subscription lifetime value has had time to compound. The report was built to say that plainly rather than dress up an early number as a finished one.预算方案本身按 90 天验证周期来设定规模，而不是当作一个无上限的开放预算，配一套统一的报表体系——访问量、转化率、客单价、分渠道的混合获客成本、订阅转化率、边际贡献——让第 30 天、第 60 天、第 90 天的检查点能直接从看板上读出来，而不是在会议室里靠争论决定。真正重要的数字是边际贡献——也就是一笔订单的收入，减去随销量变化的成本（比如货品成本、运费、支付手续费）之后剩下的部分，还没扣掉固定的日常开支——而不是毛利率，因为在获客成本还在优化、订阅的长期价值还没来得及积累的早期阶段，边际贡献通常偏薄甚至为负。报表如实呈现这一点，而不是把一个早期数字包装成一个已经成熟的结果。

What this did not include: a promised conversion rate, a promised cost per acquisition, or a revenue projection issued by Easterlies. Direct-to-consumer benchmarks for food and beverage brands exist as ranges reported by third parties, not guarantees, and the brand's own 90-day numbers were the only ones that counted. The deliverable was a credible model and a documented go, iterate, or pause decision at day 90, decided on the data the dashboard produced, not on the calendar.这次交付不包含：一个承诺的转化率、一个承诺的获客成本，或者 Easterlies 发出的营收预测。食品饮料类独立站的行业基准以第三方报告的区间形式存在，不是保证，真正算数的只有品牌自己那 90 天的数字。交付物是一个可信的模型，和一个在第 90 天有据可查的"继续、调整或暂停"的决定——这个决定读自看板产出的数据，不是读自日历翻到了第几页。