Segment size and revenue: how to find the smaller list that earns more

Every shop has had the conversation. The campaign matters, so send it to everybody. More recipients, more revenue — and in the narrow sense of one campaign's total, that is often true.

What it costs is invisible on that campaign's report and lands on the next six: complaints from people who barely remember subscribing, a worse reputation, and a slow decline in whether your mail reaches the customers who actually buy.

The arithmetic nobody does

Take a campaign sent to 40,000 that earned €6,000. That is €0.15 per recipient. Now split the audience by engagement and look again — almost every shop finds something close to this shape:

• Engaged in 90 days, 8,000 people: €4,200. €0.53 each.
• Engaged 90–365 days, 14,000 people: €1,500. €0.11 each.
• Not engaged in a year, 18,000 people: €300. €0.02 each.

The last group produced 5 % of the revenue and 45 % of the sending, and it is where essentially all of the complaints came from. The question is not whether to send to them. It is what those 18,000 sends cost you on the next campaign.

How to prove it on your own data

1. Pick a normal campaign, not a sale.
2. Send it as you always would, to everyone.
3. Afterwards, split the results by last-engagement band — 90 days, 90–365, over a year.
4. Compute revenue per recipient and complaint rate per band.

Do it twice so it is not one campaign's noise. The pattern is remarkably consistent, and seeing it on your own numbers ends the argument in a way an article cannot.

Where the real ceiling is

Cutting the dormant tail is the easy half. The harder and more valuable question is whether your engaged segment is itself being over-mailed.

• Watch revenue per recipient against send frequency. If it falls as you add a campaign per week, you have found the ceiling, and the extra campaigns are borrowing from the next ones.
• Watch unsubscribes per campaign among the engaged. Rising unsubscribes in your best segment is the most expensive signal on the dashboard and the one least often reported.
• Consider a frequency cap per person rather than per campaign — no more than N marketing messages in a rolling week, across every channel.

When a bigger list is genuinely right

There are real exceptions, and pretending otherwise is how this advice gets ignored.

1. A genuine re-engagement campaign, designed as one, sent deliberately to the dormant tail, with a plan to sunset whoever does not respond.
2. A one-off event with broad relevance — a shop closing, a major change to terms — where the message is closer to a notice than a campaign.
3. A very young list, where nobody has had time to become dormant and engagement data is too thin to segment on.

Outside those, a bigger list is usually a way of avoiding the harder work of having something worth saying to a smaller one.

Report it so the decision holds

The reason shops drift back to sending to everybody is that the monthly report shows total revenue, and total revenue goes up when you send more. Change the report.

• Revenue per recipient, per campaign, on one attribution window.
• Complaint rate per campaign, plotted, with the 0.1 % line drawn on it.
• Unsubscribes as a share of the engaged segment, not of the whole list.

With those three on the page, the conversation about widening the audience resolves itself.