A/B testing
A/B test your Instagram DMs and see what converts
See it run
A got 61 sends and 14 clicks. B got 39 and 17. Counting clicks would have crowned the loser.
Two versions of the same DM. Each send picks one at random.
How it works
Add a second version
Variant A starts as the DM you already wrote, so switching a test on never means retyping anything. Up to three arms.
It splits itself
Every send picks an arm at random. Nothing to schedule, no traffic to divide by hand.
Read the click rate
Clicks come back attributed to the version that produced them. Analytics shows each arm and marks the leader once there is enough data.
Rate, not clicks
A random split does not produce equal traffic. One arm routinely gets half as many sends as the other, and comparing raw click counts would hand the win to whichever version happened to be picked more often.
So the scoreboard is click rate. Sixty-one sends and fourteen clicks loses to thirty-nine sends and seventeen clicks, which is the correct answer and the opposite of what the totals suggest.
Only the copy changes
Buttons, links and carousel cards are shared across every arm. If the destination changed between versions you would be measuring two things at once and could not attribute the difference to either.
Under about thirty sends the result is labelled too early to call, rather than showing a leader. Four data points is not a finding.
What people use it for
Urgency vs. plain
"Here is the link" against "tap before it expires". The gap is usually bigger than people expect.
Emoji or none
Endlessly argued about, trivially testable, and audience-specific enough that nobody else's answer helps.
Short vs. explained
One line against three. Whichever wins, you stop rewriting it every launch.
Different offers
Free guide against a discount code, sent to the same keyword, judged on the same rate.
Works with
Settle it with data
Ask for it
Need something that is not here yet?
Tell us what you were hoping to find. Requests from people who are still deciding are the ones that shape what gets built next, and they go straight to the person who writes the code. Leave an email and you get an answer, not a silent roadmap.
Bigger idea, or a build of your own? The full request page has more room.