Klaviyo winback flow

The same purchase trigger as post-purchase, and a different job

A winback flow writes to someone who bought, and then did not buy again before you expected them to. It uses the same metric as a post-purchase flow. The trigger is Placed Order. The difference is the wait, and one profile filter: Placed Order zero times since starting this flow. Profile filters are checked at entry and again before each message. A person who orders during the wait fails the filter and skips the rest. A skipped count here is often the people who came back on their own.

If an ecommerce integration is connected, Klaviyo says a winback flow is already in the account. You can use it, or use one from the flow library. Building it yourself is the same trigger and the same filter, then a delay you set. Placed Order and Ordered Product are different events. The filter has to name the metric this store records for a completed purchase. A filter on the other metric skips the customers you meant to evaluate, or treats a product-level event as a reorder.

The first delay is the buying cycle, plus a little time

Klaviyo’s point about timing is the product. A couch and a bottle of shampoo do not share a repurchase window. Set the first delay a little longer than the store’s average time between orders, so you write to someone who has missed a purchase rather than to someone who is still using the last one.

If the account qualifies for predictive analytics, the method in the winback article is a segment, then an export. The segment is Placed Order at least 2 times in the last 2 years. Export it and average the Average Time Between Orders column. That average is the number you put under the delay, plus the leeway the article asks for. If predictive analytics is not available, answer two questions for this catalog: when would you want to order again, and when would you need the product again. Conditioner is their example of something you replace as soon as it runs low. A durable good can wait.

Delays add along the path. Klaviyo’s timing article uses a winback as the example: a 75-day delay before the first email and a 15-day delay before the second means the first email is about 75 days after the purchase and the second is about 90 days after it. A delay that waits until a time of day, or a day of the week, can land later than the sum. Read the elapsed day on the canvas, not the gap between two cards in isolation.

The frequency chart marks two winback sends, at 2 months and at 3 months. Those marks are the average-brand default in the library. They are two emails, where the winback article recommends three, and 2 months is not this catalog’s buying cycle until you have checked the orders. A replenishment product with a set cycle belongs in a replenishment flow. Klaviyo says to take those items out of winback. A 60-day winback email for a product people reorder every 30 days is a late second purchase, not a lapse.

Three emails, and the third one has to offer a way out

The winback article keeps the series to three emails. The first is the short one. It shows three to six products, either best sellers or new releases, and it is the place for an incentive if you use one. The second reminds them of that incentive, puts a date on it, and shows different products from the first email. It can be longer. The third is the last call: take the offer, change preferences, or unsubscribe. The unsubscribe link on that email has to be prominent. Klaviyo’s reason is deliverability as well as the legal requirement. Someone who will not come back should leave, rather than stay on the list and stop opening.

Add a delay and a message after each email, then preview each one. Review and turn on, then Live, Update status now, and save. Unconfigured messages stay draft. A draft third email means the last call never sends, and the person has already received the incentive emails with no final unsubscribe prompt from this series. The list unsubscribe may still be in the template. The article’s point is that this email is where you say it is the last one.

Apple Mail Privacy Protection can record an open the person did not make. The winback article says not to trigger a flow off opens without a way to separate those opens. A split that treats a recent open as proof of interest will treat a privacy open as interest. Clicks are the engagement the article is on firmer ground with when you branch the series.

One buying cycle does not describe every customer

Klaviyo says you can split, or build a second flow, for customers who are not on the average cycle. The conditions it names are amount spent over a period, number of orders over a period, a recent site visit, and a recent open. A person with three or more orders can see best sellers in one email and new arrivals in the next, if you are testing that. Test one change at a time. A shorter delay and a different offer in the same branch does not tell you which one moved the reorder.

If the store also sells through Buy with Prime, one winback is not enough. Klaviyo says to run a second winback on the Buy with Prime order metric, and to add Placed Order (Buy with Prime) zero times since starting this flow on the ecommerce winback. Otherwise a reorder through Buy with Prime looks like a lapse, and the ecommerce series keeps going. On Shopify, the Buy with Prime article starts that second flow from the Customer Winback template in the Buy with Prime library. Built from scratch, which is the path that article gives for other platforms, the trigger is Placed Order with the Buy with Prime icon, and the profile filters are Placed Order (Buy with Prime) zero times since starting this flow and the ecommerce platform’s Placed Order zero times since starting this flow. The example delays on that from-scratch flow are 75 days and then 15 days, and the article says to change them to the cycle you measured.

Read the delay against the orders, not against the chart

Confirm the trigger is Placed Order and the profile filter is Placed Order zero times since starting this flow, on the metric this account records. Read the first delay and compare it with the average time between orders, or with the answer you can defend for this product. Count the live messages. The article’s series is three. The chart’s series is two, at two months and three months. Only the canvas sends. Check that a replenishment flow, not this one, owns products on a set cycle. Check the Buy with Prime filter when that integration is on. Preview the third email and find the unsubscribe.

KlavyFlow places each message on the elapsed day after a read-only connection, so a winback that waits two months sits on that later day beside the creative, and a 15-day follow-up sits on its own day. It does not compute Average Time Between Orders, change the delay, or send the mail. Performance for the selected flow is a separate view over 7, 30, or 90 complete account-local days. A missing result stays missing. A winback’s first send is often outside a 7-day window. The timeline is how you see the delay that is actually configured, in each client workspace, without treating one brand’s reorder cycle as another brand’s.