Segmenting a List of Four Hundred

The advice assumes a bigger list

Segmentation is presented as basic practice: divide the list, send relevant messages to each group, improve everything.

That advice comes from organisations with lists in the tens of thousands, where a segment is still a few thousand people and the difference between segments is measurable.

At four hundred, four segments is four lists of a hundred, and a difference in response between them is a handful of people.

So the work multiplies, the results become unreadable, and the sending schedule becomes something nobody maintains.

What segmentation actually costs

• More messages to write, every time.
• More decisions about who gets what.
• Data to maintain so the groups stay accurate.
• Results too small to compare.
• A schedule that gets abandoned.

The third is the one that fails quietly: segments depend on information about people that goes out of date, and nobody updates it, so within a year the groups describe who somebody was rather than who they are.

The one split worth making

Where segmentation earns its place even on a small list, and it is usually only one.

Divide by whether the message would genuinely be different.

For most small businesses that is customers against everybody else: people who have bought from you and people who have not.

Those two groups need genuinely different messages. One knows what you are like to work with and the other does not, and a message that works for one reads oddly to the other.

That split takes minutes to maintain, since you know who your customers are, and it doubles the work rather than quadrupling it.

Anything beyond it should have to justify itself against the same test.

The test for any further split

One question, and most proposed segments fail it.

Would you write a materially different message to this group.

Not a different greeting or a swapped photograph, but different content, because their situation is different.

Segmenting by which town somebody lives in fails, unless you would say something genuinely different to each town.

Segmenting by which service somebody bought passes, if the follow-up advice differs by service.

Where the answer is that the message would be broadly the same with minor changes, send one message.

A worked example

A business with about four hundred and fifty addresses built six segments: by service, by area, and by whether somebody had bought.

Writing a monthly message became writing six, so the newsletter went from monthly to quarterly and then stopped.

Fourteen months later they restarted with two groups: customers and enquirers.

They wrote one message a month and changed two paragraphs between versions, which took about fifteen minutes more than a single message.

Response was higher than the segmented version had achieved, largely because the messages were actually being sent.

Their conclusion was that frequency had mattered more than relevance, and that segmentation had been the thing preventing frequency.

Personalisation is not segmentation

A distinction worth keeping, since the two get conflated.

Inserting somebody's name, or their town, or the service they bought, into one message is personalisation, and it costs almost nothing.

Writing separate messages for separate groups is segmentation, and it costs a message each time.

On a small list, personalisation gives you most of the relevance for a fraction of the effort.

Check it renders correctly for somebody with a missing field, since the classic failure is a message beginning with a greeting and no name, which is worse than no personalisation at all.

What to do instead at this size

Since the effort has to go somewhere and segmentation is not where it pays.

Send more often, which is the change with the largest effect on a small list and the one segmentation prevents.

Write better subject lines, which affects everybody rather than a group.

Remove people who have not opened anything in a year, which improves delivery for everybody remaining.

And ask people what they want when they sign up, which lets them segment themselves at no cost to you.

That last one is the version of segmentation that works at small scale, because the maintenance is done by the subscriber.

When it becomes worth it

Because this is a question of scale rather than principle.

Somewhere above a thousand, segments become large enough that differences between them are readable and worth acting on.

Selling genuinely different things to genuinely different audiences changes it at any size, since a trade customer and a homeowner are not one audience.

And an automated sequence based on behaviour, which segments itself without anybody writing extra messages, is worth having earlier than manual segmentation is.

Below those conditions, one message sent regularly beats four sent occasionally.

Removing people is the better lever

Worth setting against segmentation, since it is the other thing people do to a list and it returns more.

A list of four hundred where a hundred and fifty have not opened anything in a year is not a list of four hundred, and sending to all of them makes delivery worse for the two hundred and fifty who read.

Removing the inactive raises your open rate, improves how mail providers treat your domain, and costs you nothing real, since those people were not reading anyway.

Give them one message asking whether they want to stay, then remove the silent ones rather than keeping them for the sake of the number.

A smaller list that gets read is a better asset than a larger one that damages your standing every time you use it.

The counter-case

There is a version of this argument that becomes an excuse for laziness.

Sending everybody the same message when a portion of them clearly need something else is not simplicity, it is not paying attention, and people notice.

The customer-against-enquirer split in particular is not optional for most businesses, since sending a first-time introduction to somebody who has used you three times reads badly.

Some small lists are also naturally divided, such as a business serving both trade and public, where two lists are the honest structure rather than a refinement.

Make the split your business actually has, and refuse the ones that exist only in the software.

The decision

1. Count the list before deciding anything.
2. Split customers from everybody else, and stop.
3. Ask of any further split: different message.
4. Use personalisation instead of segmentation.
5. Test how it renders with a missing field.
6. Let people choose what they want at signup.
7. Spend the saved effort on sending more often.

Step seven is where the return actually is, since frequency does more for a small list than relevance and segmentation is usually what stops it.

Building the list in the first place is covered in the first hundred subscribers.

Frequently asked questions

West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.