Subscriptions

Subscription winback: bring cancelled subscribers back when they actually run out.

A subscriber who just cancelled still has product at home. Wait until it's gone, talk to the reason they left, and make restarting one click. That's most of a good winback.

A subscriber cancels on the 3rd, a week after their last 30 day pack arrived. On the 4th, the typical winback email shows up: "We miss you, here's 20% off." They still have three weeks of product in the cupboard, so they delete it. By the time they actually run out, the brand has gone quiet. A subscription winback works better when it waits for that empty cupboard, and when it speaks to the reason the person left.

This guide is about subscribers who actively cancelled. For one time buyers who just stopped ordering, our general winback flow guide is the better fit. And everything before the cancel, like save offers in the customer portal, belongs in the subscription cancellation flow.

Why subscription winback is easier than normal winback

With a normal winback you're guessing. You know someone hasn't bought in a while, and that's about it. With cancelled subscribers you know much more:

  • The exact date they cancelled.
  • What they had in their subscription and how often it shipped.
  • When their last order went out.
  • Often, the reason they gave in the cancellation survey.

Your subscription app sends these to Klaviyo as events and profile properties. Before building anything, check that the cancellation event carries the reason, and that you can see the product and interval on the profile. Our guides on Recharge, Loop, Skio and Stay AI show where to find them.

Subscription winback timing: when the product runs out

The first email should land around the time the subscriber's last order would be used up. Before that, they don't need you. After that, they may already have bought something else.

Work it out like this:

  1. Take the date of the last delivered order.
  2. Add how long that order lasts, based on pack size and interval.
  3. Subtract a few days, so the email arrives just before they run out.

In Klaviyo, the simplest version is a time delay after the cancellation event, set per product or interval with conditional splits. If you have the last order date as a property, you can be more precise with a date based flow. Either way, check your real data after a few months: when do reactivations actually happen? Move the first email closer to that.

Products that don't run out, like a monthly box or a membership, work differently. There the right moment is often the next cycle they'll miss, for example when the next box ships to everyone else.

Split by cancel reason

This is the part that makes a subscription winback work. The reason someone gave when they cancelled tells you what the email needs to say. Build a conditional split on the reason property and give each branch its own message.

Cancel reasonWhat the winback saysOffer?
Too much productRestart with a longer interval, skip any timeNo
Too expensiveThe offer, or a cheaper option like a bigger pack or slower intervalYes
Didn't see resultsHow long it usually takes, how to use it, honest and calmNo
Didn't like the productA different product or flavor, if you have one that fitsMaybe a sample
Moving or travellingRestart with a new addressNo
Switched to another brandWhat makes yours different, briefly, then leave itRarely
No reason givenWhat's new since they left, restart linkTest it

Use your own reasons. If your survey has seven options, map each to a branch. If you don't have a cancellation survey yet, that's the first thing to fix. Our cancellation survey guide covers how to set it up.

Some reasons should never get a winback. If a subscriber cancelled because their pet died, because of a health issue, or because they were unhappy with how support treated them, suppress them from the sales sequence. For pet brands we cover this in detail in the pet food subscription marketing guide.

Offer only where price was the reason

It's tempting to put a discount in every winback email. It's also expensive, and it teaches subscribers that cancelling is how you get a deal. Next time, they'll cancel on purpose.

A discount answers one reason: price. For everyone else, it misses the point. Someone with too much product doesn't need 20% off more product. They need fewer deliveries.

For the price branch, keep the offer clear and time limited, and make sure it's one your margins can carry for a few orders. Decide if the discount covers one order or several. A single cheap order followed by the full price often leads to a second cancel.

For the "no reason given" group, test whether an offer helps at all. Send half without and compare reactivations.

Make restarting one click

Every winback email needs a link that brings the subscription back with as little effort as possible. Ideally the same products, interval and address, ready to confirm.

How this works depends on your subscription app. Some can reactivate the cancelled subscription directly from a link, some need the customer to log in, and some rebuild the subscription through a pre filled cart. Find out what yours supports before you write the emails, and test the link from a real cancelled test profile. A restart link that drops people on a login page loses a good share of them.

If you can, let the restart link also change the interval. For the "too much product" branch, a link that restarts on a slower interval is the whole email.

Example subscription winback flow

  1. Just before the product runs out: Reason based email with restart link.
  2. About two weeks later: What's new since they left: new product, new flavor, a change they asked for. Restart link.
  3. Four to six weeks after that: A short plain text note from the founder. "Would you tell me what we could do better?" Replies go to a real inbox.
  4. Optional, a few months later: One last email if something genuinely new launched. Then move them to past customer campaigns.

Subject line examples, per branch:

  • Too much product: "Want it every 8 weeks instead?"
  • Price: "Your old subscription, now at 20% off for 3 orders"
  • No results: "Most people need more than three weeks"
  • No reason: "A few things changed since you left"

Only use a number like "three weeks" if it matches your product and your approved claims. The same goes for any discount: use your real offer.

What to stop doing

Starting the winback the day after the cancel. They still have product. The first email gets ignored and the later ones get less attention.

Sending the same discount to everyone. It costs margin and doesn't fix most reasons.

Keeping them in active subscriber flows. Check that renewal reminders, upsells and subscriber only campaigns stop at the cancel. A "your next box ships Friday" email after cancelling looks careless.

Sending first order offers. Exclude past subscribers from your new customer welcome discount. Someone who paid full price for months and then sees a first order offer feels foolish. Our segmentation guide shows how to build that exclusion.

Running the sequence forever. After three or four emails, stop. Move them into normal campaigns for past customers, with a lower frequency.

What to measure

  • Reactivation rate, overall and per cancel reason branch.
  • Time from cancel to reactivation. Use it to move the first email.
  • Reactivated subscribers who cancel again within two orders. High numbers here usually mean the offer did the work, not the product.
  • Revenue from reactivated subscriptions over their next few orders, compared with the cost of any offer.

Winback is one piece of the retention picture. For the full map, see our subscription churn email flow guide.

Where to startPull last quarter's cancellations with their reasons. Build a branch for your two most common reasons first, time the first email to when product runs out, and test the restart link from a cancelled test profile.

Questions we get a lot

When should a subscription winback flow start?

Around the time the last order would run out, not the day after cancelling. Work it out from the interval and pack size. For a 30 day supply cancelled right after a delivery, that's roughly three to four weeks after the cancel. Check your own reactivation data and adjust.

Should every cancelled subscriber get a discount to come back?

No. Offer a discount only to people who said price was the reason. Someone who left because they had too much product needs a longer interval, and someone who moved needs an address update. A discount for everyone costs margin and teaches people to cancel for a deal.

What is a restart link?

A link that reactivates the cancelled subscription, or rebuilds it with the same products and interval, in as few clicks as possible. Recharge, Loop, Skio and Stay AI each handle this differently, so check what your app supports before building the emails.

How many emails should a subscription winback flow have?

Usually two to four, spread over a few weeks or months. After that, move the person into your normal campaigns for past customers and stop the winback sequence.

What is the difference between a subscription winback and a normal winback flow?

A normal winback targets customers who haven't ordered in a while. A subscription winback targets people who actively cancelled, so you know when they left, what they had and often why. That lets you time and word the emails much more precisely.

Want to see where your extra revenue is?

Book a free 30 minute audit. We go through your Klaviyo together and show you what we would change first.

Get a free audit