The quiet customer is the expensive one
The retention conversation usually stops at the post-purchase sequence: the thank-you, the shipping update, the review request that lands a fortnight after delivery. That work covers the first thirty days. It does nothing for the customer who bought twice last year, liked the product, and then simply stopped showing up. They are still on your list, they still recognise your name, and they have not seen an email from you since a campaign you cannot remember sending.
This is the cheapest revenue in the store. They already paid the acquisition cost, they already trust you, and they need a reason to look again — not a pitch. The reason most stores never collect it is not that the flow is missing. It is that the flow fires on a guessed timeline, usually ninety days because that number appears in every template, and opens with a discount because the template did that too.
Your repurchase window is the trigger, not a generic 90 days
Klaviyo's own guidance is to anchor the win-back trigger to your store's average buying cycle rather than a fixed number of days. You can pull the figure in about ten minutes: build a segment of everyone who has placed at least two orders, export it, and average the Average Time Between Orders column. If you want to be stricter, look at customers who made a second purchase and find the window in which seventy to eighty-five percent of them placed it — then set the delay just after that window closes.
The number moves a lot by category, which is exactly why the template figure is wrong for most stores. Replenishable goods — coffee, supplements, skincare — often sit under forty-five days. Apparel tends to run sixty to a hundred and twenty. Furniture and other durable, high-consideration purchases run several months. A ninety-day trigger is too late for the coffee store, where a lapsed customer is already two orders deep with a competitor, and too eager for the sofa brand, where it reads as nagging.
One filter matters more than the delay: eject anyone who buys. Set the flow so it only targets customers who have placed zero orders since entering it. Without that, a customer who reorders on day two still receives the whole sequence, which is the fastest way to make a returning customer feel like they are on a list.
Escalate the message instead of opening with a coupon
Lead with a reminder, not an offer. The first email is a light, human nudge — what is new, a bestseller they might have missed, a simple acknowledgement that it has been a while. Leading with fifteen percent off trains your list to wait for a discount before they buy, and it erodes margin on the customers who would have come back on their own. Klaviyo recommends keeping the sequence to roughly three emails, and that restraint is the point: three spaced messages beat five fired over ten days, which reads as a bombardment and buys unsubscribes.
Email two, five to ten days later, goes value-led — social proof, a new arrival, a use case for the product they already own. Only the third email, usually twelve to fifteen days in, carries the incentive, and it doubles as the exit: a short, honest last call that gives them control over whether they keep hearing from you. A clean unsubscribe at that stage is a better outcome than another year of sends to someone who will never open them.
There is also a quieter rule most stores skip: exclude anyone in the win-back window from your regular campaigns. A lapsed customer receiving a flow email and a generic sale blast in the same week gets two uncoordinated asks instead of one clear reason to return.
Split the flow by value before you send anything
A thirty-five-dollar one-time buyer and a customer with a thousand dollars of lifetime value should not receive the same three emails. Sort the lapsed list by value first — recency, frequency, and spend, which is the RFM model Shopify reports on natively — and branch the flow. Your highest-value lapsed customers get the most human treatment: a perk, early access, or an actual note from someone on the team. The middle of the list gets the standard value path. Only the segment where the unit economics clearly allow it gets a monetary offer.
If your account has enough history, the same data answers the timing question better than a fixed delay. Klaviyo's predictive fields estimate a customer's Expected Date of Next Order and a churn-risk score, which lets you catch someone the week they drift past their usual cadence — while they are still warm — instead of waiting for a generic ninety-day mark. Treating at-risk and genuinely lapsed as two different moments is what separates a proactive flow from a post-mortem.
Measure repeat purchase rate, not the email
Judge the flow on repeat purchase rate and revenue per recipient, not opens. Opens tell you the subject line worked; revenue per recipient tells you whether the sequence is worth the send. Then ask the harder question: did the customers who came back buy a second time after returning? A win-back that produces exactly one order and then silence is not retention — it is a discount you paid for, and it should show up in the data as a weaker repeat rate across the cohort.
Watch the ceiling too. Past roughly twelve months of inactivity, reactivation rates fall sharply, and continuing to discount to that group costs money for little return. That group belongs in a sunset flow and eventual suppression, not in another offer. Give the whole system a quarter before you judge it — the goal is not a spike the month you switch it on, but a repeat purchase rate that quietly climbs while you get on with everything else.