When customers buy once and disappear, it’s often because nothing happens between the first order and the moment they’d need a second. To increase repeat purchase rate, measure it by first-order cohort, fix the first-order experience, time the second offer from your own reorder data, then add subscriptions, loyalty, referral, SMS or direct mail where they fit, on top of your core Klaviyo flows. If cohorts stay flat after all that, the problem is the product or the offer.
Measuring repeat purchase by cohort
The store-wide returning customer rate in most dashboards misleads. It counts anyone who has ordered before, so it rises when you cut acquisition spend and falls when you scale it, even if nothing about retention changed.
Measure by cohort instead: group customers by the month of their first order, then track what share placed a second order within 30, 90 and 180 days. Every cohort is compared at the same age, so a real improvement shows up as a line that bends.
A worked example with made-up round numbers:
| First-order month | New customers | 2nd order by day 30 | By day 90 | By day 180 |
|---|---|---|---|---|
| January | 2,000 | 6% | 14% | 20% |
| February | 2,400 | 5% | 13% | 19% |
| March | 3,000 | 4% | 11% | Too early |
| April | 2,800 | 7% | Too early | Too early |
March brought in more customers but fewer came back, so ask what changed in acquisition that month: a deeper discount, a new channel, a new hero product. April’s early lift is the first sign that a change is working.
Then cut the same metric four ways, because that’s where the answers are:
- By first product. Some entry products create repeat buyers and some don’t. This should decide what you advertise.
- By acquisition channel. Two channels with similar CAC can produce very different second-order rates.
- By first-order discount depth. Heavy welcome discounts often buy customers who never pay full price.
- By days to second order. The distribution tells you when to act.
Shopify’s built-in reports get you partway. I usually export orders with customer IDs and build the cohort table in a spreadsheet, so it can be cut any way the question needs.
The first-order experience
The second order is mostly won or lost during the first. A customer whose package arrived late, who couldn’t figure out the product, or who saw no result won’t respond to any reminder.
Check these before adding a single retention tactic:
- Delivery. Actual shipping time against what you promised, tracking updates, damaged or wrong items. Group first-time buyers’ support tickets by reason.
- First use. If the product needs setup, a routine or a few weeks before it works, say so in the box and in the first post-purchase messages. Some one-time buyers simply used it wrong or quit too early.
- The unboxing. Packaging is the one channel every customer sees. Use the insert to explain how to get the best result and what to try next. A QR code to a 60-second how-to beats a paragraph of brand story.
- A sample of the likely second product. If your data shows buyers of product A tend to buy product B next, put a sample of B in the box.
- Expectations set by ads. If the ad promises results in a week and the product takes a month, the customer judges it a failure on day eight.
When I audit DTC brands, the quickest retention wins usually come from this list, not from new flows.
Timing the second purchase
Most brands send the “come back” message on a schedule someone picked. Use reorder data instead:
- Take every customer with two or more orders and calculate the days between their first and second order.
- Plot that distribution for each main product or category. Note the median and the point by which most second orders have happened.
- Start reminders a little before the median, and move customers to win-back once they pass the later point.
For consumables, the usage window drives timing. As an example, if a bottle lasts about 60 days and the median reorder lands around day 52, the first reminder belongs around day 42-45, leaving time to ship before they run out.
For non-consumables, timing is about the next logical product. Build a simple matrix with first-order products in the rows and second-order products in the columns. It tells you what to recommend next, and often surfaces a pairing nobody expected.
Klaviyo’s predictive analytics can estimate each customer’s expected next order date once your account has enough order history. It’s a useful input, but check it against your own distribution.
The common mistake is a discount on day 3 to “lock in” the second order. It pulls forward purchases that would have happened at full price and trains customers to wait for codes.
Subscriptions and replenishment
A subscription turns the second order from a decision into a default. It fits when the product is used up on a predictable schedule, the customer doesn’t want to choose again each time, and running out is a nuisance. It fits poorly for high-variety, irregular-use or trend-driven products.
The design choices matter more than the discount:
- Default frequency from data. Set the default interval from your reorder distribution, not a round 30 days. Subscribers who pile up unused product cancel.
- Flexible controls. Skip, delay, swap and change frequency. A subscriber who can push a delivery back two weeks often stays; one whose only option is cancel leaves. Keep cancellation simple too: several markets regulate auto-renewal, and a hidden cancel button turns into chargebacks and complaints.
- When to offer it. Offer it at checkout, and again to customers who just placed a second one-off order. They’ve already shown the habit, so the ask is easier.
- A modest incentive. A small subscribe discount or free shipping. Check it against margin, because it applies to every future order.
On Shopify, subscriptions run through Shopify’s own Subscriptions app or third-party apps, so the platform is rarely the constraint. Track subscriber retention by month since signup and read cancellation reasons. “Too much product” means the frequency is wrong, not the product.
Loyalty and referral programs
Loyalty
Most points programs reward customers who were going to buy anyway. To move repeat purchase rate, design around the first-to-second order gap, usually the biggest drop in any cohort table.
- Award credit on the first order that can only be spent on the next one, with an expiry that lands inside your reorder window.
- Prefer perks that don’t cut margin on every order: free shipping, early access, members-only products.
- Add tiers only if your best customers buy often enough to reach them.
Test incrementality before you scale. Members usually look better than non-members because your best customers are the ones who join. Hold out a random share of eligible customers, or compare cohorts from before and after launch.
Referral
Referral programs amplify word of mouth that already exists; they rarely create it. Ask at a moment of success, such as a second order or a five-star review, not on a first order’s confirmation page. A give-get structure (credit for both the friend and the customer) is standard, and the customer’s credit doubles as a reason for their own next order.
Retention channels beyond email
Email usually carries most of the retention load, but inboxes are crowded. Other channels earn their place when they have a specific job:
| Channel | Best job | Fits when | Watch out for |
|---|---|---|---|
| SMS | Reorder reminders, shipping updates, early access | Short buying cycles, mobile-first customers | Needs separate explicit consent; costs more per send; fatigues fast |
| Direct mail | Postcards at the reorder window or for win-back | Higher order values, customers ignoring email | Cost per piece; measure with a holdout |
| Paid social to customers | Launches and cross-sell to existing buyers | New products, a clear next purchase | Takes credit for orders that would have happened anyway |
| Packaging inserts | Next product, how-to, referral code | Every brand | Generic content nobody reads |
Don’t mirror every email in SMS; keep it for the few messages where timing matters. On paid social, Meta lets you define existing customers from customer lists or purchase data, but its settings for splitting spend between new and existing buyers change often, so check what your campaigns currently offer. Decide deliberately how much retention spend belongs in ads at all.
When retention is a product problem
Flows, points and SMS can move a customer who liked the first order toward a second one. They can’t make someone want a product that disappointed them. Signs the problem sits upstream:
- Second-order rates stay flat across cohorts after flow, timing and first-order fixes ship
- One entry product has healthy repeat rates and the others have almost none
- Reviews and return reasons mention quality, results or fit rather than price
- Customers acquired with deep discounts rarely come back at full price
- Repeat buyers only return during sales
- One-time buyers say they switched to a cheaper or better alternative
To find out which applies, survey one-time buyers who are past your reorder window and ask, in open text, why they haven’t reordered. A handful of honest replies often explains more than a dashboard.
Then match the fix to the cause:
- Weak entry product: stop advertising it as a first purchase and lead with the product that creates repeat buyers.
- Discount-driven acquisition: reduce first-order discount depth and judge channels on 90-day cohort value, not first-order ROAS.
- Quality or results problem: fix the product before spending another dollar on retention.
- One-time product by nature, such as a mattress or a durable tool: stop chasing repeat rate and focus on order value and referrals. See how to increase average order value.
This is the diagnostic I start with in ecommerce growth engagements, reading retention, acquisition and product data together, because the fix for low repeat rates often sits in the ad account or the product line, not the email tool.
Get it built
If most of your customers buy once and disappear, I can build the cohort data, find where the second order is lost and prioritize the fixes. The Growth Audit is $1,500 fixed and credited if we continue. See pricing or get in touch.