Split testing your Discord outreach

A test is only useful if one thing changes. Two entirely different messages sent to two different niches tells you nothing about either.

Change one variable

• The opening line
• The offer itself
• The ask at the end
• The destination page

Give it enough volume

Splitting 1,000 DMs into two arms of 500 will rarely separate two decent messages. Test at 5,000 or more per arm when the decision matters, and treat small runs as sanity checks only.

Use separate links

Each arm needs its own invite or tracked URL. Without separate links you cannot attribute joins, and reply counts alone will mislead you.

Keep a log

Record the date, servers, quantity and result for every run. Six months in, that log is the most valuable marketing asset you own.

Keep reading