The right email for the right people

Send each group what fits it. The same email to everyone is the most common reason people stop reading, and the easiest to fix.

The short version

• Groups double clicks. Mailchimp found segmented campaigns drew 100.95% more clicks and 9.37% fewer unsubscribes than the same senders' campaigns to everyone (Mailchimp's own customers, updated 2017).
• Start with three groups: customers and prospects; engaged and quiet; and one interest you ask about.
• Ask people what they want. Groups people chose themselves did best of all: 74.53% more clicks and 25.65% fewer unsubscribes.
• Judge by clicks, replies and purchases, never by opens alone. Apple Mail opens email on its own, so an "open" no longer means someone read it.
• Use a first name sparingly, and only when the names are clean.
• Send time matters less than people think. Choose a weekday morning in your readers' time zone, keep it, and test by clicks and sales.
• Every subscriber stays in at least one regular send, unless you mean to rest or remove them.

Why groups work

A group is worth having only if it changes what you'd say. Mailchimp compared about 11,000 segmented campaigns with the same users' campaigns to everyone:

And by what the group was built from:

The strongest groups come from what people told you they want.

The groups to start with

Add them in this order:

1. 1

   Customers and prospects. People who've bought need different email (how to use it, what goes with it, thanks) from people still deciding.

2. 2

   Engaged and quiet. People who clicked, replied or bought recently get your full rhythm. The quiet get less, then a win-back. This protects your inbox placement for everyone.

3. 3

   Interest. What they chose at sign-up, which gift they came for, what they click.

4. 4

   Purchase history, for a shop: best customers, new customers, customers drifting away.

5. 5

   Place and time zone, for a local business or a list across time zones.

6. 6

   Where they joined, when different forms or events bring people who want different things.

A list under about a thousand people gains little from fine slicing. Two or three of these are enough, and the work goes into the email itself.

Ask people

The best data you can hold is what people tell you:

• At sign-up, one optional choice.
• In the welcome sequence, one question with three or four answers that match what you sell: "Which of these sounds most like you?" Each click tells you which group they belong in.
• Later, a way to change their answer.

In the EU and the UK, collect only what you'll use, and say why when you ask.

What people do beats who they are

What people click, what they look at, the date of their last order and a reply predict what they want far better than their age or job title. An email sent because of one action is the extreme case, and it's why sequences earn so much more per email: Why email works. Use details like a buyer's role, a size or a region only where they change the offer.

Opens no longer mean reading

Since 2021, Apple Mail can open an email's pictures in advance, which records an "open" whether or not anyone read it. So:

• "Engaged" means clicked, replied or bought in the last 30, 60, 90 or 180 days, chosen by how often you send. Treat opens as a weak extra signal.
• A group of "openers" quietly includes people who never read anything, and a group built on opens alone can silently drop everyone whose mail app hides them.
• Once a month, check that nobody has fallen out of every send by accident.

A first name, sparingly

In field experiments by Sahni, Wheeler and Chintagunta (Marketing Science, 2018), a name in the subject line lifted opens by about 20% (from about 9% to about 11%), sales leads by 31%, and cut unsubscribes by 17%. It works because a name stands out, so it likely wears off if every email uses it. "Hi FIRSTNAME" or a misspelled name does more harm than none.

People expect email that fits them: in McKinsey's 2021 survey, 71% said they expect it, and 76% said they're frustrated when it doesn't happen. For a small business that means the basics: the right product for a customer, nothing sold to someone who just bought it, the right town for an event.

Personalise in layers, cheapest first: the right group gets the email at all (most of the gain), then sections that change by group, then recommendations from what they looked at or bought, then names and details where the data is reliable.

When to send

The big studies disagree on the best hour, which tells you the hour isn't the big lever:

What they agree on: weekday mornings in the reader's time zone are safe, evenings can be good for clicks and buying, and Saturday is weakest. What matters more is a regular day, so you're expected, and timing that fits the email: a lunch offer before lunch, a B2B newsletter on a weekday morning, a weekend sale on Thursday or Friday.

Test a time by clicks and sales, never opens: Apple's mail servers may fetch an email when it arrives, not when it's read.

Personal data has rules

Grouping people by what they do with your email is ordinary practice. Health, religion, sexuality, ethnicity, politics and children are sensitive: they need explicit consent in the EU and the UK and care everywhere, and guessing them from clicks still counts as using them. Keep every group explainable. If someone asked "why did I get this?", the honest answer should be one they'd accept. More in Email marketing and the law.

Doing it here

• A campaign can go to one list or several, and each list can be narrowed by any column from your upload, such as a town or "customer: yes": Choose who gets it and get it ready to send.
• From a campaign that's gone out, you can save the people who clicked, the people who opened without clicking, or the people who didn't open, as a segment, and send each group what fits next.
• The report counts real people and leaves machines out, because security scanners open email and follow links on their own: Real people, not machines.
• The Daily Sending Window sets the hours your mail goes out. A big send drips out at a natural pace for each mailbox provider, so the time you pick is when sending starts, not the minute every copy lands: Can I schedule a campaign for later?