Loyalty · Email timing

When should a replenishment email send?

By Mirow · October 3, 2026 · 7 min read

Send a replenishment email shortly before a plausible reorder window for the product and quantity the customer bought. Use purchase history to choose a timing test, then check whether the reminder gets additional customers to reorder. A 30-day delay across the whole store is a weak starting point when one customer bought a single pack, another bought three, and a third bought something that will last for years.

Start with one product, one pack size and one reminder. The method below is Mirow's recommended testing approach. The numbers are hypothetical, not client results or a benchmark.

Measure the next purchase of the same product

Build a cohort: a group of customers who made a comparable purchase during a defined period. For an initial replenishment test, use first buyers of one consumable product in one quantity. Separate subscription orders, multipacks and gifts where those are reliably identified. Record the time from each customer's first qualifying order to their first subsequent order of the same product.

An order for a different product does not prove the first item ran out. Decide in advance whether another flavor or size counts as a replacement; use that same definition in the analysis, flow filters and outcome measurement. Exclude canceled initial orders. Count only completed, paid repeat orders; exclude test orders and free replacements. Decide in advance how full and partial refunds affect a qualifying repeat, and apply that rule consistently.

Choose a follow-up window long enough for the product cycle, and give everyone the same opportunity to return. If the analysis window is 90 days, include only customers whose initial orders are at least 90 days old and count repeats within each customer's first 90 days. A buyer acquired last week cannot yet be classified as a 90-day non-repeater. Check separate acquisition periods for stockouts, promotions or seasonal changes before pooling them.

Klaviyo's replenishment guidance recommends adjusting timing to customers' established buying cycles. The next step is to inspect the distribution, rather than copying a delay from a template.

Read the reorder window without dropping non-returners

Suppose 200 first-time buyers each purchased one pack of the same consumable. All have a complete 90-day observation window. The table counts each buyer once, at their first same-product reorder; 140 have no recorded same-product reorder within that window.

Scroll horizontally to see all columns.

Among the 60 buyers who return, 24 ÷ 60 = 40% reorder by day 30. Across the original cohort, 24 ÷ 200 = 12% reorder by day 30. By day 45 those figures are 80% of returners and 24% of all buyers. Both descriptions are correct, but they answer different questions.

The median reorder time among the 60 returners falls in days 31–45. These grouped counts cannot identify the exact day. They also cannot establish a median for the whole cohort: only 30% have reordered by day 90. The other 70% may buy later, buy elsewhere or never need the product again.

For this hypothetical cohort, days 21 and 28 are reasonable candidates to test before the days 31–45 interval. The unequal interval widths mean the table cannot establish a daily ordering peak. They are choices for an experiment, not estimated optimal send dates. Existing promotions or reminders may already have shaped the purchase history, and placing an order does not reveal the exact day a customer ran out.

If you want to use future repeat purchases to support acquisition spending, that requires a separate CAC payback calculation based on observed repeat orders. A replenishment schedule alone does not establish future customer value.

Keep the timing tied to the purchase

Write down the event that starts the clock. An order-to-order interval already describes when customers placed their next order under the previous delivery conditions. Do not automatically subtract shipping time from it again. A supply estimate measured from delivery is different: it needs an explicit allowance for the customer's response time and the replacement shipment. Keep those assumptions separate from observed reorder data.

Use product context to decide whether a replenishment message belongs in the journey at all.

Scroll horizontally to see all columns.

The copy should carry the same uncertainty as the timing. “Ready for another pack?” offers a useful action. “You must be running out” claims knowledge the order history does not provide. Link to the correct product and quantity, and check availability before recommending it.

Make the flow stop asking after a reorder

For a product-specific Klaviyo flow, scope both the purchase trigger and the subsequent-purchase check to that product. Klaviyo's setup instructions describe a profile filter that checks for purchases since flow entry before each email. A store-wide purchase filter could suppress a coffee reminder just because the customer bought a mug.

Klaviyo checks profile filters before flow actions; failing the check skips that action. Do not assume a new order instantly removes the customer from every queued step. Test the actual event properties and filters with a fresh buyer, a same-product repeat buyer and a buyer of a different product. Also check subscription exclusions, refunds and whether repeated triggers could create overlapping reminders.

Replenishment is a promotional message in Klaviyo's transactional-email guidance. Keep it within the customer's marketing-email permissions and preserve unsubscribe and suppression handling. Klaviyo recommends express consent; an order record by itself is not your permission check.

Test whether the reminder earns the send

For the hypothetical single-pack product, assign eligible, subscribed customers at the initial qualifying order to three groups: one reminder on day 21, the same reminder on day 28, or no replenishment reminder through day 90. Randomize before either send date and keep assignment fixed; prevent later orders from assigning the same customer to another group. Klaviyo supports random splits for testing flow timing.

Keep the product, copy, offer and other marketing treatment comparable. Start without a discount so the test isolates timing. Stop the reminder when a qualifying reorder has already happened, and honor changed permissions in every group. Keep those skipped customers in their originally assigned groups when calculating outcomes; removing them would remove some early repeat purchasers from the comparison.

Use store orders to calculate buyers with a qualifying reorder by day 90 ÷ all customers originally assigned to that group. Report the counts beside the rates. Track unsubscribes and complaints with explicit denominators, and include net sales after refunds as supporting context. Where actual costs are available, compare contribution dollars per assigned customer after product, fulfillment, payment and messaging costs. Do not call revenue profit.

The no-reminder group estimates what happens without this particular message. Compare the full window, not just purchases immediately after an email: an earlier reminder may move an order forward without increasing repeat purchasing by day 90. Email-platform attribution alone cannot resolve that distinction.

Before launch, set the smallest improvement worth acting on and the sample needed to distinguish it. If volume cannot support three groups, test one candidate date against the holdout first. Do not declare a winner from a handful of orders or stop when one arm first looks better. Wait until the planned sample has its full follow-up window; an uncertain result is a reason to keep learning, not to add more reminders.

For the next working session, export one mature product cohort, retain the non-returners and name two plausible send dates. Then document the trigger, repeat-purchase exclusion, holdout and decision date before turning the flow on.