ConvoHut
Retention use case

Give a quiet customer a real reason to come back.

Reactivating a customer who already trusts you is cheaper than acquiring a stranger. Most win-back programmes fail not because the customer left, but because the message assumes the only reason to return is a discount.

Free PDF. No email required.

At a glance

TriggerNo purchase or activity beyond your threshold
GoalReactivate the customer
Primary KPIReactivation rate
Messages2–3
Estimated setup timeUnder 25 minutes

What is a win-back campaign?

A win-back campaign targets customers who have stopped buying or engaging beyond a defined period. On WhatsApp it works as a short two-way conversation: acknowledge the gap, give a specific reason to return based on what they bought before, and let the customer tell you why they left.

Key takeaways

  • Set the inactivity threshold from your actual purchase cycle, not a round number.
  • Segment by past value before you decide what the offer is worth.
  • A win-back that asks why they left is worth more than one that converts silently.
  • Two messages, then stop. Persistence here damages the brand.

Why this leaks revenue today

01

The threshold is guesswork

Ninety days means something different for coffee than for furniture. A generic window messages the wrong people at the wrong time.

02

Everyone gets the same discount

Your highest-value lapsed customers get the same 10% as someone who bought once, two years ago, on sale.

03

Nobody learns why they left

One-way win-back campaigns recover a little revenue and zero information. The second one performs no better.

The signal ConvoHut acts on

The journey only starts when all of these are true. Getting the entry conditions right matters more than the message.

No purchase for 90+ daysNo engagement across channelsPreviously purchased 2+ timesLifetime value above threshold

What has to be connected

Purchase historyLast order date, frequency, categories, lifetime value
Product cycleExpected repurchase window for what they actually bought
EngagementWhether they are ignoring you everywhere or only on one channel
Offer rulesWhat you are willing to give, and to whom

The journey

Built once, then it runs unattended. Every branch has a defined outcome, and the journey stops the moment the goal is met.

Trigger

Inactive beyond threshold

Condition

Lifetime value check

Message

Send personalised return message

Delay

Wait 5 days

Then it branches on what the customer does

Replies with a reason

Route to a human — this is the valuable outcome

Clicks through

Straight to the relevant category, not the homepage

No reply

One final message, then suppress for 6 months

Goal: Order placed or reason captured — journey stops

A journey without a stop condition is a journey that gets your number blocked.

The messages, and why each one exists

Starting points, not scripts. You rewrite them in your own voice and approve every message before it goes live.

Day 0

Asks rather than sells. The reply is often more valuable than the order.

It's been a while since your last {category} order. Anything change, or just busy?

Day 5

Gives a reason to return that is about the product, not a discount.

If it helps, here's what's new in {category} since you last looked: {links}

Day 12

Ends cleanly and offers the exit. This protects the channel and your sender reputation.

Last one from us — we'll stop here. If you'd rather not hear from us at all, just say stop.

What it looks like to the customer

This is the part a broadcast tool cannot do. The customer replies, the question gets answered from real data, and the journey continues — in one thread.

If the AI reaches the edge of what it has been approved to answer, it escalates to your team with the full history attached. The customer sees one continuous conversation.

It's been a while since your last coffee order. Anything change, or just busy?
Switched to a subscription elsewhere honestly
Fair enough — good to know. We do subscriptions now too, with pause any time. Want me to show you the closest match to your old blend?
Go on then
Here it is. Same roast profile, fortnightly delivery, cancel whenever.

Illustrative conversation. Not a customer transcript.

What to measure

Pick one primary metric before launch and agree how it is calculated. Everything else is secondary.

Reactivation rate

Lapsed customers who place an order inside the agreed window.

Recovered revenue

Revenue attributed to reactivated customers.

Reason captured

How many customers told you why they left. Track this deliberately.

Opt-out rate

The health check. A rising opt-out rate means the threshold or the tone is wrong.

How we keep this honest. The attribution window is agreed before launch. Direct and influenced conversions are reported separately. Where volume allows we hold out a control group. We do not claim every order that happens after a message.

Set it up

  1. 1

    1. Define inactivity from your own data

    Use the actual purchase cycle for each category, not a single site-wide number.

  2. 2

    2. Segment by past value

    Decide which lapsed customers justify an offer and which get a product-led message.

  3. 3

    3. Write a message that asks

    The question is the point. A win-back that only sells learns nothing.

  4. 4

    4. Set the suppression window

    Anyone who does not respond should not be contacted again for at least six months.

  5. 5

    5. Launch on one segment

    Start with high-value lapsed customers where the message can be most specific.

Then tune it

Move the threshold before you change the offer

Most underperforming win-backs are mistimed rather than under-incentivised.

Test product news against discount

For repeat-purchase categories, what is new often beats what is cheaper.

Read the replies weekly

The reasons customers give here are the highest-quality retention research you will get for free.

Questions this raises

How long should someone be inactive before a win-back?

It depends on the purchase cycle. Consumables might justify 60 days; considered goods might need 12 months. Derive it from your own repeat-purchase data rather than a benchmark.

Should I lead with a discount?

Not by default. Discounting every lapsed customer trains the profitable ones to lapse deliberately. Reserve offers for segments where the maths supports it.

Isn't WhatsApp too intrusive for a lapsed customer?

It is more intrusive than email, which is exactly why the frequency has to be lower and the message better. Two messages, then stop, and always offer the exit.

What if they reply that they are unhappy?

That routes to a human immediately. A complaint recovered well is worth more than a discounted reorder.

How do I avoid messaging people who are still active elsewhere?

Check engagement across channels, not just purchases. Someone reading every email but not buying is a different problem.

What counts as a successful win-back?

Agree it before launch. Usually an order inside the attribution window, but capturing the reason for churn is a legitimate secondary goal.

Related use cases

Free download

The Win-back Playbook, as a PDF.

Everything on this page in a document you can send to your team or your client: entry conditions, journey logic, the messages with the reasoning behind each one, the metrics and the setup steps.

No email required. Take it and build the journey somewhere else if you want to — the logic is the useful part.

Win-back Playbook

PDF · 4 pages · Version 1.0

Download the playbook See all resources

Your customers are already showing intent.

Turn those moments into automated conversations that move customers toward conversion. We work with a small group of early access customers and set the first journey up with you.

Join the waitlist