Short answer: most stores have built a machine for the first purchase and no machine for the second. All the budget, all the analytics and all the attention go into acquisition, and once the parcel ships the customer simply drops out of view. That is the central marketing gap in small and mid-sized ecommerce — and also the cheapest one to close.
Why this is the biggest lever
The arithmetic is simple and unflattering. To get a new customer you pay for ads, compete for the click, fight the distrust of an unfamiliar store and lose 70% at checkout. To get a second purchase from an existing customer you need an email and a reason.
The second effect is resilience. A store living purely on new customers is fully exposed to ad prices: the auction gets more expensive, profit drops. A store with 30–40% repeat purchases survives rising traffic costs, because part of its revenue isn't re-bought every month.
The third effect is that you can afford to outbid competitors. If your customer is worth €180 over two years rather than €60, you win any auction where the other side only counts the first transaction.
The four retention numbers
Calculate them once, then review quarterly:
- Repeat customer share — how many customers placed more than one order. Measure over at least a year, otherwise the figure is understated.
- Second-purchase window — how many days on average until the second order. This is the calendar for every reminder you send.
- Purchase frequency per year — how often an active customer buys.
- LTV — how much margin (not revenue) a customer produces over their lifetime. Compare it against CAC.
The most useful of these is the second. If the average second purchase lands on day 45, an email on day 30 meets live intent, while an email on day 120 is chasing a train that already left.
What to do, in order of payback
1. A post-purchase sequence that serves rather than sells. The most-opened emails in ecommerce are about the order. Confirmation, dispatch, tracking, "it arrived — is everything alright?". Put usefulness into them: how to use it, how to care for it, what to do if there's a problem. These are emails people wait for, and they shape the relationship more than any advertising.
2. Something inside the parcel. Unboxing is the one moment where you have 100% of the customer's attention. A card with a thank you, a short guide and a personal code for the next order costs pennies and works better than banners.
3. Cycle-based reminders for replenishables. If the product runs out in 60 days, an email on day 50 isn't spam, it's a service. For stores selling consumables, cosmetics, pet food or filters, this is usually the most profitable automation you can run.
4. Segmentation instead of blasting everyone. Even a simple RFM scheme (recency / frequency / monetary) yields four clear segments: new, active, dormant, best. Each has its own reason to receive an email. One campaign to the whole list is how you burn the list.
5. Reactivating dormant customers. People who haven't bought in 6–12 months are the cheapest source of orders after active ones. One honest email ("it's been a while, here's what's new") with a segmented offer brings a share of them back.
6. A range that supports a second order. Accessories, consumables, bundles, subscriptions. If you have nothing to sell them a second time, the problem isn't marketing, it's the product range — and that is a strategic question, not an email task.
Owning a channel versus renting traffic
The core idea of this topic: social platforms and ads are rented traffic; your customer list is an asset you own. The algorithm changes, an account gets suspended, the bid goes up — and reach disappears overnight. An email and SMS list stays with you.
Which leads to a practical conclusion: capture the contact before the purchase happens, and do it honestly.
- One signup field with a clear benefit ("we'll tell you when your size is back") outperforms a "−10% for subscribing" pop-up
- An address obtained by trickery unsubscribes from the first email and damages your deliverability
- Consent has to be valid under GDPR — otherwise the asset becomes a liability
Write rarely and with substance. A store that emails once a fortnight with something genuinely useful gets read. A store that sends three promotions a week lands in the Promotions tab and then in spam.
Loyalty programmes: when they work and when they are just a discount
They work when:
- The product is bought regularly (consumables, cosmetics, coffee, pet food)
- Rewards are reachable in a sensible time — 3–4 purchases, not 20
- The rules are understandable in ten seconds
They don't work when:
- Purchases are rare (furniture, appliances) — the customer forgets before the second one
- The programme is a permanent discount in disguise and simply erodes margin
- Enrolling is harder than buying
What stores lack from a marketing standpoint: the short list
If the whole topic is boiled down to the gaps I see most often:
- No reason to come back. The store is indistinguishable from ten others except on price — and a price war is won by whoever has the deeper warehouse.
- Nothing to say. The only content is "sale", so only discount hunters read the emails.
- No segments. Everyone gets the same message, which makes it irrelevant for most of them.
- Discounts instead of value. Discounts train people to wait for discounts. Useful content, service and delivery speed don't.
- No content that answers pre-purchase questions. Guides, comparisons, buying advice — the material that brings people from search and AI assistants and makes you recognisable before the transaction.
- No margin measured per customer. Without LTV you can't know what you can afford to pay for acquisition, so every decision is a guess.
Developer note
- Export orders with per-customer history. "How many orders does customer N have" should be answerable without dumping the whole database into a spreadsheet.
- One customer identifier everywhere. Guest orders by email must merge with accounts, otherwise LTV can't be calculated.
- Don't run segmentation through WP-cron. On a store with a large list it is unreliable; move it to a system job or a queue.
- Keep transactional email on its own channel. Order confirmations should not share a domain and reputation with campaigns: marketing landing in spam must not drag delivery notifications with it.
- Data handling has to be real. Consent, unsubscribes and deletion requests must work technically, not as "we'll delete it manually".
- Integrate the CRM/ESP through events, not exports. An order status change should move the customer between segments immediately.
Practical check
- Calculated the repeat customer share over the last 12 months
- I know the average window to the second purchase
- Calculated LTV in margin terms and compared it to CAC
- A post-purchase email sequence exists, not just "order received"
- The parcel contains something besides the invoice
- At least four list segments with different messages
- A dormant-customer reactivation flow is running
- Emails are collected honestly and with valid consent
- There is an answer to "why should the customer come back to us" beyond price
What's next
The final topic of the guide is about what makes every improvement above either possible or impossible: the state of the store itself. Plugins and tech debt: the real cost of ownership.