How Small Is Too Small: The Minimum Segment Size Worth Mailing
A segment of 40 people is not a segment. It is a list, and the results it produces are mostly noise.
A segment with 40 people in it is not a segment. It is a list of 40 people, and mailing them as a group tells you nothing about whether the message worked. Segments under 500 profiles lack the statistical power to show a real difference in open or click rates, and a campaign sent to a tiny segment produces a number that is mostly noise. Keep campaign segments in the thousands, and save micro segments for flows, where the message is triggered and the audience is not the point.
The noise problem, in numbers
Take a segment of 40 people. Send a campaign. Two of them open it. The open rate is 5 percent. Send the same campaign to a different 40 people and five open it, and the rate is 12.5 percent. The difference between the two campaigns is three people, and the rate moved by seven points. That is not a finding. That is randomness, and any decision based on it is a guess dressed up as data.
The threshold where the numbers start to mean something is around 500 profiles. Below that, a single open or click moves the rate by whole percentage points. Above it, the rate stabilizes enough to compare one campaign against another. This is not a rule anyone invented. It is arithmetic, and it applies to every segment in every store.
Campaign segments belong in the thousands
The practical rule is to keep campaign segments in the thousands. A segment of 3,000 profiles with a 20 percent open rate means 600 opens, and a 2 point lift means 60 more opens, which is a real difference. A segment of 300 with the same rate means 60 opens, and a 2 point lift means 6 more opens, which could be a Tuesday.
This does not mean small segments are useless. It means they are not for campaigns. A segment of 200 VIP customers is worth mailing, because the point is to reach those 200 people, not to learn something from the result. The mistake is treating a campaign to a small segment as a test, and then acting on the result as if it proved something.
Micro segments belong in flows
The place micro segments work is flows, where the message is triggered by behavior and the audience size does not matter. A flow that sends a reminder to the 40 people who abandoned a cart today is not a test. It is a response to an action, and the trigger is the point, not the sample. The flow works whether 40 people or 400 people enter it.
The distinction is between a message that is sent to a group and a message that is sent because of something the person did. Campaigns are the first kind, and they need a large enough group to read the result. Flows are the second kind, and they work at any size, because the result is the recovery, not the rate.
The audit that finds the tiny segments
Most stores have dozens of segments they never use, and many of them are tiny. The audit is simple: list every segment, with its size and the date it was last used. Any segment under 500 profiles that has not been used in a campaign in 90 days is a candidate for deletion. Any segment under 100 profiles that has never been used is a candidate for deletion today.
Deleting a segment does not delete the people in it. They stay in the main list, and they stay in the flows that target them by behavior. What disappears is the clutter: a segment list with 30 entries instead of 200, and a campaign workflow that starts from segments worth mailing instead of segments worth forgetting.
Keep reading
Related guides on this site: segment decay and why stale segments cost you opens segments nobody activates and the audit that finds them segment overlap and when one buyer sits in three segments More from Creative that adapts to every customer segment.
Frequently asked questions
What is a segment that is too small?
A segment that is too small to show a real difference in performance. With fewer than 500 profiles, a single open or click moves the rate by whole percentage points, so the numbers are noise, not signal.
What is the minimum size for a campaign segment?
About 500 profiles is the floor for a campaign segment. Below that, the results are not reliable enough to act on. Keep campaign segments in the thousands when you can.
Are micro segments ever useful?
Yes, but only in flows, where the message is triggered by behavior and the audience size does not matter. A flow that sends to 40 people who abandoned a cart is fine, because the trigger is the point, not the sample.
Can a small segment ever show a real result?
Yes, if the segment is large enough and the difference is real. A segment of 5,000 with a 10 percent lift is a finding. A segment of 200 with a 10 percent lift is two people.