Email and customers · Lesson 5.6
Measuring a Win Back Campaign: Metrics That Matter
After you send a win back campaign, the temptation is to look at whatever your email tool shows first. That is usually opens and clicks. They are easy to read and often misleading. This post sets out the few numbers that tell you whether a win back campaign worked, and how to read them in a short weekly check.
No benchmarks are given, because they depend on your product, list and offer. We are not going to invent a "good" rate. Lesson 5.6 walks you through it step by step: Win back customers who went quiet.
Judge a win back campaign by orders and replies first, then unsubscribes and complaints. Treat opens as a weak hint.
The five win back campaign numbers
| Metric | How to get it | What it tells you | Trust level |
|---|---|---|---|
| Orders from the group | Order list filtered to the group, since the first send | Whether people came back | High |
| Replies | Count by hand, read each | Why they left, in their words | High |
| Unsubscribes | Email tool | Whether the message felt unwelcome | High |
| Spam complaints | Email tool or Postmaster Tools | Whether you are over reaching | High |
| Opens and clicks | Email tool | A rough hint of interest | Low |
Notice what the table leaves out: revenue per email, "engagement score", and so on. They may have a place later. For a first campaign, five numbers are enough to decide what to change.
Why opens mislead
Apple's own Mail Privacy Protection page says the feature is meant to prevent senders from learning information about a person's Mail activity, such as when and how many times an email was opened. It says Protect Mail Activity downloads remote content in the background by default, regardless of whether the person engages with the email.
Open tracking usually relies on remote content loading. So for people who use Apple Mail with this protection, an open can be recorded when the person never read the email. A win back list is full of people who stopped paying attention, which makes inflated opens more confusing. If your open rate looks good but orders are flat, believe the orders.
Clicks are somewhat better than opens, since they need an action, but some security software also clicks links. Orders and replies are the cleanest.
Orders from the group
Take the group you emailed, which you wrote down before sending, and count how many placed an order after the first send. Divide by the group size if you want a percentage, and keep the raw count too.
A few cautions.
- Some would have ordered anyway. Some people in the quiet group were about to come back. Without a comparison group you cannot separate the two. If your list is large enough, you can hold back a small random slice and email nobody in it, then compare. For small lists, accept the uncertainty and say so when you review.
- Check the window. Count orders in a fixed period, such as until a week after the last email.
- Split the groups. Repeat buyers and one time buyers often behave differently. Mailchimp suggests treating them separately.
Replies
Read every reply. A reply such as "I moved away" tells you that part of your "quiet" list was never reachable. A reply such as "I did not know you added that range" tells you what to say next time. Replies are small in number but rich. Tag each as: wants to order, no longer needs it, price, service issue, other. A few tags will show a pattern.
Unsubscribes and complaints
Unsubscribes are not failures. Klaviyo's winback guide says it is better for someone to unsubscribe than to ignore your emails or mark them as spam. Treat a modest number of unsubscribes as the list cleaning itself.
Complaints are different. Gmail's sender guidelines say to keep spam rates reported in Postmaster Tools below 0.3%. Yahoo's sender requirements say the same figure. If you cross that on a win back send, you are emailing too many people who do not want it. The fix is a smaller, better chosen list next time.
If you cannot see complaint data in your tool, you can check your reply folder for angry messages and your unsubscribe volume. Small senders often will not have Postmaster Tools data, so treat missing data as unknown, not as zero.
A weekly read, in ten minutes
Set a recurring time, such as Friday, during the campaign and the week after it.
- Count ordersFrom the group, since the first send. Write the raw number.
- Read repliesTag each one. Reply to anyone who asked a question.
- Check unsubscribesNote the number and whether it jumped after a specific email.
- Check complaintsIf your tool reports them. Note any.
- Glance at opens and clicksOnly to spot a broken link or a subject line that obviously failed.
- Write one line"This week I learned ___. Next time I will change ___."
That last line is the point. Measuring is only useful if it leads to one change.
Comparing two versions fairly
If you want to test two versions, such as a subject line or an offer, split the group randomly in half, keep everything else the same, and compare orders. Do this only when each half is big enough that a few orders do not decide the result. With a small list, a test often shows noise. In that case it is better to run the same campaign twice over several months and compare, or just keep a record and look for patterns.
Signs it is working, and signs it is not
You cannot define "working" without your own baseline, but you can check direction.
- Orders from the group are more than zero, and replies are mostly friendly: keep going, adjust one thing.
- Many unsubscribes but few orders: the list or the message was off. Revisit who you chose.
- Many complaints: send to fewer people, check your unsubscribe link, and slow down.
- Almost no replies and no orders: the window may be wrong, or the offer too weak. Change one thing only.
Related reading
For the process these numbers belong to, see the WIN framework and the complete guide. For errors that spoil your numbers, see 5 win back campaign mistakes.
Questions people ask
What is a good win back rate?
We cannot give a number. It depends on your product, list and offer. Track your own over several rounds.
Should I use open rate at all?
Only as a weak hint. Apple's Mail Privacy Protection can inflate opens, so rely on orders and replies.
What about revenue?
You can track the revenue of orders from the group, but count orders first. Revenue can swing on one large order.
Where to go next
The last step of Lesson 5.6, Win back customers who went quiet is seeing who came back and deciding what to try next. The lesson also covers finding people who stopped buying, writing the note in your voice and sending a small offer to the right people only.