Your cheapest revenue is a customer who already bought once

Every order from a new customer carries the cost of finding them: ad spend, agency fees, the discount that got them over the line, and the traffic that never converted but still appeared on the invoice. That cost resets with each new person. A repeat order from someone already in your database carries almost none of it — the trust exists, the shipping address exists, often the saved payment method exists too. So retention is not a channel sitting politely next to paid acquisition; it decides whether paid acquisition is affordable at all. If a customer is worth exactly one order, you can only ever spend what one order earns. If the average customer buys three times, your ceiling on acquisition cost roughly triples, and every competitor bidding on the same keywords with a single-order model is fighting you with a smaller budget.

Cart abandonment is the sharpest signal in the store: the shopper chose a product, chose a variant, and stopped. Baymard Institute's aggregated cart-abandonment research puts the average across studies near 70%, and while much of that is browsing that was never going to convert, a real share is recoverable friction — an unexpected shipping cost, a distraction, a payment method they wanted to check first. A reminder sent within an hour catches the session while intent is still warm; a second one the next day catches the deliberate delayers. Browse abandonment sits one step earlier and converts less, so it deserves a lighter touch: three views of the same product without an add-to-cart is a signal, one pageview is noise. Both flows depend on knowing who the anonymous visitor is, which is a storefront problem before it is an email problem.

• First cart reminder within the hour, second the following day
• No discount in the first reminder — you would buy back orders you already had
• Trigger browse abandonment on repeat views, not a single visit
• Suppress the whole flow the moment the order lands

The message a customer is most likely to open is the one about the order they just placed. Shipping and delivery updates get read because they are useful, which makes the days after a purchase the best-attended window you will ever have, and the worst one to waste on a generic newsletter. Use it to reduce anxiety first — where the parcel is, what to do if something is wrong — then to earn the next step: care instructions, the accessory that pairs with what they bought, a review request timed to after the product actually arrived. Replenishment is the same idea with a clock on it. If a consumable lasts about six weeks, the reminder belongs in week five, not on the first of the month. Win-back sits at the far end: a repeat customer who has gone quiet for two of their own cycles deserves one honest email, not a monthly campaign that trains them to ignore you.

A modal that covers the screen two seconds after a first-time visitor arrives is an interruption tax paid on every session to collect a small number of low-quality addresses. The alternative is not abolishing the popup; it is asking at a moment when the visitor already has a reason to say yes. Back-in-stock and price-drop alerts collect people telling you exactly what they want. A consent checkbox at checkout collects buyers, who are worth more than subscribers. A genuinely useful sizing guide, compatibility checker or care sheet collects intent. Then the welcome flow has material to work with: it can pick up the product they were actually looking at, instead of opening with a blanket 10% code that teaches every new subscriber to wait for a discount before they buy anything.

• Back-in-stock and price-drop alerts: intent, not interruption
• Ask at checkout, where consent is a natural step
• Trigger the popup on exit or second visit, never on arrival
• Let the welcome flow reference what they browsed

Age bracket and country tell you very little about whether someone buys again. Behaviour tells you nearly everything: what they bought, how long ago, how often, at what price point, and whether they only ever order during a sale. A customer with three full-price orders and a customer with one order redeemed against a 20% code are not the same audience and should not receive the same email. The working version of this is unglamorous — recency, frequency, value, plus category affinity — and it needs no data science project, only clean order and event data leaving the store in a usable shape. That is where analytics that can attribute a repeat order stops being a reporting nicety and starts deciding who receives what, and how often.

Sending is the part owners ignore until the mail stops arriving. Three DNS records carry most of the weight: SPF declares which servers may send for your domain, DKIM signs each message so it cannot be quietly altered in transit, and DMARC tells receiving servers what to do when the first two fail. Google and Yahoo's bulk sender requirements, introduced in 2024, made all three effectively mandatory at volume, alongside one-click unsubscribe and keeping spam complaints below their published threshold. Send marketing from a dedicated subdomain, warm it up instead of blasting a cold list, and never buy addresses. Consent matters legally and mechanically: in the EU, GDPR and the ePrivacy rules expect an opt-in you can evidence, while in the US, CAN-SPAM permits more but still demands a working unsubscribe and an accurate sender identity and postal address.

• SPF, DKIM and DMARC published for the sending domain
• Marketing on its own subdomain, warmed up, never a purchased list
• One-click unsubscribe honoured in hours, not next campaign
• A stored consent record: timestamp, source, exact wording shown

None of these flows run on copy alone; they run on events the storefront has to emit. An abandoned-cart email needs to know a cart existed and no order followed. Replenishment needs the order's contents and date. Identity has to survive the gap between a phone browsing at lunch and a laptop buying at night, or your best segments quietly split into strangers. Here the boundary is worth stating plainly. WooCommerce keeps what it is good at: products, stock, orders, coupons, payments, refunds — and the transactional order emails it already sends. Those receipts are a different system from marketing email, with different rules; a receipt does not require marketing consent, and a campaign must not ride along inside one. What changes is the storefront: reliable event tracking, stable identity across sessions and devices, and reorder and account flows fast enough that a click from an email lands somewhere usable.

If your catalogue is a one-off purchase with a ten-year replacement cycle, the honest strategy is referral, accessories and reviews — not a replenishment flow with nothing to remind anyone about. If first orders arrive late or damaged, a win-back campaign asks customers to forgive an operations problem you have not fixed yet, and it will underperform for exactly that reason. And below a few hundred orders a month, the cheaper first move is usually one flow, abandoned cart, configured properly in whatever tool you already pay for, rather than a retention programme and the engagement that comes with it. Do the arithmetic before the design: monthly orders, a realistic recovery rate, a realistic order value. If the answer is small, spend the money on the product or on acquisition, and come back when it isn't.