On a Tuesday evening, Dana ordered two bags of coffee from a small roaster she had found through a friend's Instagram story. The order came to $64, and it was her first purchase from the brand. Over the next 97 days, the roaster sent her eleven messages. Every one of them went out on schedule, yet none of them brought her back for a second order. Her inbox shows a problem that comes up again and again in post-purchase engagement ecommerce: the messages were sent on time, but none of them knew what the others had said.
| Day | What Dana received | Sent by |
|---|---|---|
| 0 | Order confirmation | Store platform |
| 1 | Shipping notification with a tracking link | Store platform |
| 4 | "Delivery exception" alert | Carrier |
| 5 | Reply to her "where's my coffee?" email, 26 hours after she wrote in | Helpdesk |
| 6 | Ticket closed once the bags arrived two days late | Helpdesk |
| 8 | "20% off sitewide, this weekend only" | Email platform |
| 21 | Monthly newsletter about a new single-origin coffee | Email platform |
| 38 | "How did we do?" review request | Reviews app |
| 60 | "We miss you! Here's 10% off" | Email platform |
| 75 | "Still thinking about us? Take 15% off" | Email platform |
| 97 | "Last chance: your 20% expires tonight" | Email platform |
The delivery ran two days late, so Dana had to email support to find out where her order was. The helpdesk fixed the problem, but the email platform never found out about it. Two days later, it sent her a sitewide sale as if nothing had gone wrong. Her coffee ran out around the third week, which was the best moment to invite her to reorder, and she got a general newsletter instead. By the time the first win-back email arrived on day 60, she had most likely found another roaster. Attentive's 2026 State of Loyalty & Retention survey found that 88% of shoppers had bought from a brand that was new to them in the previous three months, so other brands had plenty of time to reach her first.
The win-back emails performed about as well as such emails usually do. Klaviyo's 2026 benchmark data, based on more than 110,000 brands, puts the average click rate for win-back emails at 2.8%. Emails sent soon after a purchase get 5.1%. The difference comes down to timing, because by day 60 most buyers have already made up their minds.
This guide is for founders, heads of ecommerce, and retention and CX leads who want more first-time buyers to come back without giving away more margin. It covers:
why most win-back emails start too late
the four reasons buyers stop buying, and the early signs of each
what to send at each stage between the order and day 90
when a sequence should pause, change or stop
what more returning buyers are worth in dollars, and what the usual tools cost
when scheduled emails are enough and when an AI agent helps
the five numbers that show whether win-back is working
Why does a win-back campaign usually start after the buyer has already left?
Post-purchase engagement covers every message a store sends between the order confirmation and the second order. Good post-purchase messages are timed to what actually happens: the order ships, the delivery runs late, the product arrives, or the buyer is about to run out. A win-back sequence is the final part of this journey. It only starts once a buyer has gone longer than usual without ordering again. Most stores build the win-back sequence carefully and leave the earlier messages to their default settings.
The timing explains why that approach struggles. Industry estimates put the typical time to a second order at around 15 to 35 days for most product categories. The usual win-back campaign ecommerce teams set up, however, starts at day 60, 90, or 120. By then, the period when buyers are most likely to reorder has already passed.
Buyers also pay the most attention right after they order. In The Second Purchase Is Where E-Commerce Is Actually Won, Ephanti's Founder's Corner points to the 48 to 96 hours after an order as the most engaged moment in the customer journey. Order tracking emails are opened 50 to 80% of the time in that window, while a typical marketing campaign is opened 15 to 20% of the time. Revenue follows the same pattern. Klaviyo's 2026 benchmarks show emails sent after a purchase earning $1.80 per person on average, compared with $0.82 for win-back emails.
Teams often group three different kinds of email under "retention." The table below separates them, because each one needs different information to work.
| Post-purchase messages | Win-back sequence | List re-engagement | |
|---|---|---|---|
| Who it's for | Every buyer, from the day they order | Buyers who have gone longer than usual without a second order | Subscribers who stopped opening emails |
| When it starts | At order confirmation | When the buyer passes the store's normal gap between orders | After a set period with no opens |
| What sets it off | Real events like shipping, delays, delivery, and running out | Time since the last order | No activity on the email list |
| What it needs to know | What they bought, delivery status, open support tickets | Full history, including complaints and past discount use | Email activity only |
| What success looks like | A second order in the first few weeks | A returning buyer who keeps ordering | A cleaner email list |
Our take: We think most brands put their effort in the wrong place. They spend weeks polishing a day-90 discount email while the first three weeks after delivery run on carrier updates and a general newsletter. The cheapest win-back is the one you never have to send, because the buyer reordered while the product was still in their kitchen.
Four reasons buyers stop buying, and why forgetting is only one of them
When a buyer doesn't return, most teams assume they forgot about the brand and send a discount to remind them. Some buyers do forget, but there are three other common reasons. Each one shows up in the store's data weeks before the win-back email goes out, and each one needs a different response.
They forgot, because nothing gave them a reason to return
Some buyers drift away simply because nothing after delivery gave them a reason to come back. In an August 2026 Shopify Community discussion about second orders, merchants said the product experience and the post-purchase experience mattered more than email campaigns or loyalty programs. One merchant made a sharp point: buyers rarely leave because they forgot, and more often because not coming back was easier. A one-click reorder link, sent around the time the product runs out, removes that friction far better than another reminder email.
Something went wrong, and the store kept selling anyway
Some buyers leave because a problem was fixed but never acknowledged by the rest of the business. That is what happened to Dana. Support closed her ticket, and two days later marketing sent her a sale. Delivery problems are a common cause, as The WISMO Tax explains, and buyers care about reliability more than many brands assume. In Locus's Q2 2026 survey of US online shoppers, 34% said fast delivery mattered most, while 56% put reliability or easy returns first.
How a brand responds after a problem matters even more. The same Founder's Corner piece cites research showing that companies with connected customer engagement keep 89% of their customers, while companies with disconnected engagement keep 33%. Ephanti calls this problem digital amnesia in Why Ephanti Exists. Each tool remembers its own part of the story, so the business as a whole forgets what the customer went through.
The product didn't work the way they expected
Some buyers never got full value from what they bought, because nobody answered a simple question after delivery. A coffee buyer who uses the wrong grind for her machine decides the beans taste bitter. A shopper whose jacket runs small decides the brand isn't for her. Neither of them complains, so neither shows up in a support report. Some of these buyers send the product back, which we cover in how returns and refunds shape the next order, and the rest simply go quiet.
A short, helpful message a few days after delivery fixes much of this. Ephanti's own analysis found that first-time buyers who receive personalised post-purchase messages from the brand are 45% more likely to make a second purchase.
Another brand reached them first
Some buyers leave because a competitor got their attention while the store was waiting for day 60. Attentive's 2026 research found that although 88% of shoppers tried a new brand last quarter, 77% still shop regularly with five brands or fewer. That leaves only a few regular spots in each buyer's routine, and brands compete to keep theirs. Buyers can also tell when a brand is focused elsewhere. In GetResponse's 2026 customer loyalty research, 67.3% of customers said brands value new buyers more than loyal ones.
Our take: We believe "they forgot" remains the go-to explanation because it is the only one an email calendar can act on. The other three reasons show up in delivery updates, support tickets, and product questions, and the email tool never sees any of them. So brands keep sending reminders to buyers who really needed an apology, an answer or a better reason to stay.
What should post-purchase engagement ecommerce teams send between day 0 and day 90?
It is easier to plan these messages around four stages than around a fixed number of emails. Each stage has its own trigger, its own job, and its own facts the system must know before it sends anything. The table below sets out the four stages. You can also use each row as a test for your current tools.
| Stage | What sets it off | What the message should do | Best channel | What the system must know |
|---|---|---|---|---|
| Order to delivery (first 48 to 96 hours) | Order placed, shipped, or delayed | Confirm the order, flag any delay before the buyer asks, share a short guide to getting started | WhatsApp or SMS for status, email for the guide | Carrier status and the promised delivery date |
| Two to three days after delivery | Delivery confirmed | Ask how the product is working, answer questions, suggest one item that goes with it | A channel the buyer can reply on | What they bought and any open support ticket |
| When the product is likely to run out | Expected run-out date, or the store's usual gap before a second order | Send a one-tap reorder link, or let them know an item they wanted is back in stock | The channel the buyer uses most | Quantity bought, how fast it gets used, live stock |
| When the buyer goes quiet | No second order after 1.5 to 2 times the store's usual gap | Start the win-back sequence, leading with something useful and saving any offer for last | The channel the buyer has replied on before | Full history, including complaints and past discounts |
The rule in the last row follows common retention practice. It sets the trigger from the store's own gap between orders, not from a round number borrowed from another brand. The three stages above it exist so that fewer buyers ever reach that point. Together, they move the repeat purchase rate, which is the share of first-time buyers who place a second order within a set period.
Three quick tests show whether a tool can run these stages or can only send reminders:
The delay test. On day 8, can it tell a buyer whose delivery was late apart from a happy buyer, and send each of them something different?
The reply test. If a buyer replies to the day-3 check-in with a question, does the reply reach someone, or something, that can answer it in the same conversation?
The stop test. If the buyer reorders through WhatsApp instead of clicking the email link, does the sequence stop on its own?
Ephanti is built to work this way. On Ephanti's E-Commerce page, the loyalty and repeat stage includes post-purchase follow-ups, repeat-purchase outreach, win-back sequences, loyalty engagement and review requests, all running on one customer record. Before an agent contacts a shopper, MEVA, the engine behind Ephanti, already has their cart, order history, loyalty tier, last support ticket, and preferred channel in front of it.
When should the sequence pause, change its message, or stop?
Most win-back guides explain timing and subject lines in detail. Very few explain when a sequence should hold back, even though these rules protect the most customers. They depend on information from outside the email tool, which is why so many stores don't have them.
Three rules cover most situations:
Pause while a problem is open. If a buyer has an open ticket, a pending refund or a delayed order, promotional messages should wait until the issue is fixed and acknowledged. Under this rule, Dana would never have received the day-8 sale.
Change the message after a bad experience. When a good customer has a bad experience, the next message should be an apology, a fix or a goodwill gesture, not the next promotion on the calendar. On Ephanti's Retain & Expand page, unhappy customers spotted through conversational surveys are routed to a recovery flow automatically, while happy customers are invited to refer a friend.
Stop as soon as the buyer comes back, on any channel. A buyer who reorders over WhatsApp or by phone should not get a "last chance" discount email the next morning. That only works when every channel updates the same customer record.
Discounts need their own rule. Klaviyo's 2026 benchmark report warns that offering discounts too readily teaches shoppers to wait for one before buying. For that reason, most retention teams save the offer for later messages and send it only to buyers who ignored the earlier ones. A useful rule of thumb from retention practitioners is that if more than 60% of returning buyers used the discount code, the program has built a discount habit rather than real loyalty.
Sending to everyone has a cost too. A win-back campaign ecommerce brands send to every inactive address, including buyers with open complaints, wastes discounts on people who would have returned anyway. It also teaches the rest of the list that going quiet gets rewarded.
Our take: We think the most valuable rule in any retention program is "don't send this yet." Any tool can schedule a message. Very few can hold one back because of something the customer told another team last week, and that restraint is often what keeps the customer.
What does a higher repeat purchase rate add to your store's revenue?
Retention budgets are easier to approve when the value of each returning buyer is shown in dollars. The example below uses one store, and each benchmark is linked so the numbers can be checked.
The example store is a US direct-to-consumer brand that gains 12,000 first-time buyers a year, with an average order of $70. These store figures are our assumptions, and the benchmarks applied to them are linked.
| Step | Benchmark or assumption | Result |
|---|---|---|
| Buyers who return within 12 months today | 25%, the low end of the industry estimate of 25 to 30% | 3,000 second orders |
| Value of each extra percentage point | 120 more returning buyers × $70 | $8,400 a year |
| Moving from 25% to 30% | 600 more returning buyers × $70 | $42,000 a year |
| Cost of finding 600 new buyers instead | $75 per customer, the figure used in Ephanti's Founder's Corner example | $45,000 in acquisition spend |
These figures count only the second order. Buyers who come back once tend to keep coming back, so the real value grows over time.
When the messages are sent, the result changes as well. If all 12,000 new buyers receive post-purchase messages earning Klaviyo's 2026 average of $1.80 per person, those messages bring in about $21,600. If the roughly 9,000 buyers who never return are left to a win-back sequence earning the average $0.82 per person, that brings in about $7,380. The top 10% of brands reach $2.77 and $1.60, respectively. These are benchmark ranges, not guarantees, but they show that most of the value in post-purchase engagement ecommerce comes early in the journey.
What the usual retention tools cost
Most stores spread these messages across several tools, and each tool has its own pricing. The table below prices a common setup for this store at public US list prices from 2026. Vendor names appear only in the source links.
| Tool category | What the store gets | Monthly cost |
|---|---|---|
| Email and SMS platform | About 25,000 contacts (email around $400) plus 10,000 text messages ($125) | About $525 |
| Loyalty and rewards app | Mid tier with VIP levels, up to 2,500 orders a month | $199 |
| Reviews app | Mid tier, up to 1,500 orders a month | $119 |
| Helpdesk with AI agent | 2,000 tickets and 190 automated replies, billed monthly | $550 |
| Total | about $1,393 a month, or roughly $16,700 a year |
Not included: international text rates, WhatsApp message fees, extra tickets or AI replies, extra integrations and custom enterprise pricing. Annual billing lowers some of these costs, while a busy sales season raises several of them.
The higher cost never appears on an invoice. The support ticket sits in the helpdesk, the order history sits in the store platform, the loyalty tier sits in the loyalty app, and the win-back sequence runs in the email tool. No single tool knows enough to pause, change or stop a message. Customer Engagement Is Breaking in the Same Five Ways traces this problem back to its root, which is tools that don't share what they know.
According to Ephanti's E-Commerce page, typical customers replace 4 to 6 separate tools within 60 days. They keep Shopify, their payment provider and their warehouse in place, and Ephanti replaces the support and outreach tools that sit on top. Ephanti bills on the interactions it handles, with unlimited users and no per-seat fees. It doesn't publish prices, so the best way to compare is to enter your own numbers in the E-Commerce ROI calculator.
A full platform isn't the right first step for every store. If you sell products people buy rarely, such as furniture or high-end electronics, a simple email sequence with a delivery check-in and occasional new-arrival updates may be enough for now.
Scheduled emails or a customer retention AI agent: which one does your store actually need?
Most win-back programs today run as what email tools call a "flow." A flow is a set series of automated messages that starts when something happens, such as 60 days without an order, and then sends each email on a fixed schedule. It sends the same messages to everyone in the group, and it does that reliably. An AI agent works differently. It knows the buyer's history, can hold a two-way conversation, and can act on what the buyer says. The table compares the two.
| Scheduled email flow | Customer retention AI agent | |
|---|---|---|
| What starts it | A date or a list rule | Any real event, such as a delay, a delivery, a likely run-out or a reply |
| What it knows | What the email tool stores | Order history, delivery status, support tickets, loyalty tier and preferred channel |
| When the buyer replies | The reply waits in an inbox | The agent answers in the same conversation |
| What it can do | Send a link | Set up a reorder, process an exchange within policy or update an address, and record it in the store |
| When something goes wrong | It keeps sending unless someone steps in | It pauses or changes the message |
| When it's enough | Few replies, long gaps between orders, simple products | Products that run out, clothing with fit questions, buyers who reply often or use WhatsApp |
The difference matters most when a buyer has a question rather than needing a reminder. Email vs. AI Agents: What Actually Recovers Abandoned Carts in 2026 explains why email hits a limit whenever a shopper needs an answer, and the same limit applies after the purchase. If Dana had replied "these beans taste bitter in my French press," the roaster would have had its best chance to keep her. A flow can't use that reply. An agent with her order in front of it can suggest a coarser grind and offer the right coffee for her next bag.
The channel matters too, because many buyers reply on WhatsApp far more readily than on email. Ephanti's WhatsApp AI Agent handles these conversations in the thread the buyer already uses. The Unified Inbox keeps WhatsApp, email, and voice in one conversation, so buyers never have to repeat themselves when they switch. Because an agent that takes action must be accountable, every action an Ephanti agent takes is logged and can be reviewed, as the Trust Center explains, and refunds or goodwill credits above a set limit wait for a person to approve them.
You don't have to choose one or the other. Many stores keep a simple flow for buyers who never engage and use a customer retention AI agent for the moments when buyers reply, ask questions, or need something done. That usually means the weeks between delivery and the first reorder.
Our take: We don't think flows are the problem, because they do exactly what they were built to do. The problem is expecting a fixed schedule to handle a conversation. Once a buyer replies, what they need is someone, or something, that understands their order.
Five numbers that show whether your win-back is working or just discounting
Most retention dashboards show opens, clicks, and revenue for each email flow. Those numbers can look healthy even when the program is teaching buyers to wait for a discount. The five numbers below give a more honest picture, and you can pull each of them from the order data you already have.
Second-order rate by month of first purchase. Track how many of each month's new buyers order again within 30, 90, and 365 days. The 30-day figure shows whether your early messages are working long before the yearly figure moves.
Typical days to a second order. If this number is growing, your early messages are arriving too late or saying too little. Your win-back trigger should move with it.
Returning buyers who didn't need a discount. Count the buyers who came back without using a win-back code. If most of them needed the code, you are paying for their return instead of earning it.
Third orders from buyers you won back. A buyer who returns once with a discount and then disappears again was never really won back.
Messages held back because of an open problem. Count the promotions you paused because of a ticket, a refund, or a delay. If this number is zero, your program has no pause rule, and some of your win-back emails are reaching unhappy customers.
A simple way to start is to take last quarter's first-time buyers, split them by whether they contacted support, and compare how many from each group came back. The E-Commerce Playbook covers how to plan these steps across the first 90 days and which common mistakes to avoid.
Dana's inbox, the second time around
Now imagine the same 97 days with one system holding Dana's whole story. On day 3, a WhatsApp message tells her the carrier has flagged a delay and gives her a new delivery date, before she thinks to check. When the bags arrive on day 6, the next message apologises for the wait and includes a short grind guide for her French press. The weekend sale on day 8 never reaches her, because her delivery problem is still recent. On day 19, a message reminds her she's probably running low and links to her usual bags, already in her cart. She reorders the next morning without a discount, and the day-60 win-back sequence never starts because she never went quiet.
The roaster's team begins that week with almost nothing waiting for them, apart from one goodwill credit above the approval limit that needs a person to sign off. That is what post-purchase engagement ecommerce should look like: fewer messages, each one aware of what came before it, and a second order that arrives while the first is still fresh.
As Sri Sabesan, Founder & CEO of Ephanti, puts it: "The second purchase is where e-commerce is actually won."
Bring your ugliest workflow. Name the job nobody on your team wants to do, whether it's the win-back list nobody trusts or the reorder that keeps slipping through, and we'll show you Ephanti finishing it in your systems, on the channel your customers actually use. Book a demo.
Frequently asked questions
When should a win-back campaign for an ecommerce store start?
A win-back campaign for ecommerce stores should start at about 1.5 to 2 times the store's usual gap between orders, not at a fixed 60 or 90 days. It works much better when earlier messages have already covered delivery, first use and the likely reorder date.
Should win-back emails include a discount?
Not in the first email. Start with something useful, such as new arrivals, a tip or a reorder link. Save the discount for buyers who didn't respond to the earlier emails, and keep track of how many returning buyers only came back for the code.
What is a good repeat purchase rate for an online store?
Industry estimates put the typical 12-month repeat purchase rate at around 25 to 30%. Products people use up, such as food and supplements, often score higher, while electronics and furniture score lower. Compare yourself with your own category and track the rate by month of first purchase.
What's the difference between post-purchase engagement and win-back?
In post-purchase engagement ecommerce, post-purchase messages run from the order confirmation to the second order and are timed to real events like delivery and run-out dates. Win-back is the last part of that journey, and it starts only once a buyer has gone longer than usual without ordering.
Can a customer retention AI agent replace my email flows?
A customer retention AI agent works best alongside your flows rather than replacing them. It handles the moments when buyers reply, ask questions, or need something done, such as a reorder or an exchange, while a simple flow keeps reminding buyers who never engage.
Does Ephanti handle win-back and repeat purchases?
Yes. Ephanti's E-Commerce solution includes post-purchase follow-ups, repeat-purchase outreach, win-back sequences, loyalty engagement and review requests. These run across WhatsApp, web chat, email, Instagram, Messenger, SMS and voice on one customer record, and the platform works with Shopify, Shopify Plus, WooCommerce and BigCommerce.