What is a segment?#
A rule that selects part of your list at the moment you send. Everybody on a newsletter list is in several segments at once, which is the whole difference between a segment and a list: the reader is not filed into one box, the reader is the answer to several questions at the same time.
That also means a segment is not a copy of anybody. It is a filter over the records you already hold, so a person who unsubscribes leaves every segment at once rather than needing removing from six of them.
How do I segment an email list?#
- Write the send first and decide what it should not say to. That is the segment, and it arrives from the message rather than from a menu.
- Name the one fact that separates the two halves. A tag they carry, a thing they did, a date they arrived, a place they are.
- Check the count. A segment of eleven people is a personal note, not a segment, and a segment of everybody is not a decision.
- Send to that half and keep the same subject line you would have used for the whole list, so the comparison means something.
- Leave the rule in place, because the next send uses it against a list that has changed without telling you.
The order matters. Senders who begin from the filter end up with fourteen segments describing a list of four hundred people, and then spend the time maintaining categories instead of writing the issue.
What are good examples of email segmentation?#
Six, in the order a small sender should build them. Each one answers a question you can already answer about most of your names, and none of them needs a data project.
| The segment | How you define it | What you send that half |
|---|---|---|
| Bought from you | A tag applied at the sale, or a source that means a customer | The next thing they buy, not the thing they have already bought |
| Read you recently | Opened or clicked in the last 60 to 90 days | The normal cadence, and the asks |
| Gone quiet for a year | No open across the sends you can count | One re-permission email rather than the whole series |
| Asked for a topic | The tag they chose when they joined | Only the issues about that thing, which is why they gave the address |
| One location or one site | A tag from the form, the event or the delivery address | Dates, closures and local news that a second office does not want |
| Joined through one source | Where the address came from, kept as a tag | A welcome written for what they signed up to |
| A role address | info@, admin@, office@ in the email column | Either the once-a-month note or nothing at all |
Which segment should I build first?#
The engaged half, because it is measurable from data you already have and it solves the two problems a growing list creates at once. You get somewhere to send the frequent thing, and you get a name for the part of the list that is not reading, which stops you concluding that the newsletter has stopped working.
The second one to build is the topic someone chose on the way in, because that is the segmentation the reader did for you. Somebody who ticked the workshops box is not asking for shipping news, and the address arrived with the reason attached.
Should I segment by behaviour or by what people told me?#
Both, and they answer different questions. What somebody told you is intent, and it is the better guide to what a first email should be about. What somebody did is evidence, and it is the better guide to whether to keep sending at all.
Behaviour is the one that goes wrong, because it depends on tracking that is only half true. An open is a picture being fetched, and image blocking, preview panes and privacy proxies all move that number without a reader deciding anything. Use behaviour to stop mailing people rather than to decide what to sell them.
What is the difference between a segment and a separate list?#
| The decision | A segment | A second list |
|---|---|---|
| A person on both | Counts once, and one send can reach them through either rule | Counts twice, and gets mailed twice |
| They unsubscribe | Leaves every segment at the same moment | Has to be honoured in each list, and often is not |
| What you are billed for | Usually nothing extra, since the names already exist | Often the same contact counted twice |
| Reporting | You have to ask the question each time | Clean per list, which is the reason people do it |
| When it breaks | A rule stops matching and a send goes to nobody | A reader gets two emails and one of them gets reported |
The honest reason separate lists keep winning is that they are easier to see. A list is a thing with a name on a screen and a segment is a filter you have to remember. If the second audience needs its own unsubscribe, its own sender and its own consent, it is not a segment and should not be built as one.
How do I segment an email list in Mailchimp?#
The exact screen belongs to them and their interface moves, so take the path from their own help page rather than from a blog post. What transfers from any tool is the model: tags applied to people, a filter built from those tags and from engagement, a count shown before you commit, and an audience that can be mailed through a rule rather than through a copy.
Two details to check in whatever tool you land on. Whether the segment is saved as a named thing you can reuse, since a filter you rebuild by hand each month will eventually be rebuilt wrong. And whether a scheduled send keeps the rule or freezes the list of names, because those two behave very differently when somebody unsubscribes between now and Tuesday.
Do segments cost more?#
In a tool that bills stored contacts, no: a segment is a view over names you are already paying for. In a tool that bills emails sent, the only extra cost is the extra sends you make possible, which is a decision rather than a fee.
What costs money is the number of campaigns a plan allows rather than the number of segments. A plan capped at so many sends a month makes segmentation expensive in practice, because every split is another campaign from the same budget. What email marketing costs is where that arithmetic gets worked.
What happens when a segment matches nobody?#
The send should refuse to go and tell you it found nobody, which is the friendliest thing a tool can do at that moment. A campaign mailed to an empty audience is scheduled, marked as done, and reported as a zero, and the reason is invisible until someone asks why nothing arrived.
A segment matching nobody usually means a tag stopped being applied, a filter got saved with a stale date, or the rule expects a field that imports leave empty. That is why the count belongs on the send screen and not buried in an audience view you open twice a year.
How often should segment membership change?#
Continuously, and that is the point of a rule instead of a folder. An engaged segment is a rolling window, so names come in when they read and leave when they stop. A topic segment is closer to permanent, because it records what somebody asked for.
The one that needs a human is the quiet segment. A rule that says no opens in the last year will keep a name in it for the rest of its natural life unless you decide what happens next, and the sensible answer is a single re-permission email and then a removal. How to clean an email list covers that pass.
Should new subscribers get a different email?#
Yes, and this is the segmentation nearly every sender gets right eventually. The first two weeks after somebody joins is when they are most likely to read anything you send, and a welcome written for a stranger beats the archive dropped on them a month later. Welcome email examples covers the letter, and what an autoresponder is covers the machinery that sends it without you.
How do I stop one reader getting five emails a week?#
Cap the total rather than the segment. A reader inside three overlapping audiences receives one email per campaign, and if the campaigns are separate the only protection is a rule about how often a single person can be mailed. This is where frequency is decided rather than discovered. How often should you send is the decision, and segmentation makes it visible.
Do I need tags before I can segment?#
For the behaviour segments, no, since opens and clicks come from the sending records rather than from anything you type. For everything about a person, yes, and the tag has to be applied at the moment it becomes true: on the form, at the sale, when the ticket is booked.
A tag decided afterwards is a task nobody finishes, which is why the source of the address is worth capturing on the way in. How to grow an email list is where that field pays for itself, because it tells you which door produced the readers and which produced the bounces.
What is the mistake most senders make with segments?#
Splitting for a reason that flatters the sender. The segment exists because the message should differ, and if the email would be identical to both halves, the split was a way of feeling organised.
The second mistake is reading a small number as a signal. Mail a segment of sixty people and the difference between two opens and nine is a coincidence, so a split that halves your audience also halves what you can learn from it. What a good open rate is is where that limit gets discussed properly.
Build a segment from tags, opens and clicks on the send screen and see the count before it goes, in Campaign Kicker. The rule travels with a scheduled campaign, an unsubscribe leaves every segment at once, and a send that matches nobody is refused rather than reported as a zero. The free plan mails 5,000 emails a month with no card.
Split the audience