Experiments

A/B test your captions. Each variant goes out as a real post with its own tracked link, and the winner is decided on revenue. Needs revenue attribution on.

Experiments let you test caption wording with real posts. You give one idea; seenpaid writes a few variants; each goes out as a real post; the results decide the winner.

How it works

1. Create the test

   Give an idea, a platform, and how many variants you want (2 to 8, default 3). seenpaid drafts them. Nothing is live yet.

2. Launch it

   Every variant publishes as its own post.

3. Let the numbers decide

   The dashboard shows each variant with its results and marks the leader.

4. Reuse the winner

   Take the winning caption into the composer and scale it: repost it, recycle it, or use it as the template for the next batch. Experiments live at seenpaid.com/experiments and in your agent’s tools.

What decides the winner

• With revenue attribution on, each variant carries its own tracked link and the winner is the variant that earned the most. Ties go to clicks.
• With attribution off, variants publish as plain posts with no tracked links, so there are no clicks or sales to compare and no winner can be called. Turn attribution on before launching a test. See Turn on revenue attribution.

Because every variant is its own post with its own link, attribution is clean: a sale can only belong to the variant whose link the buyer clicked.

Driving it from an agent

• create_experiment: idea + platform + variant count.
• launch_experiment: publish every variant.
• list_experiments / get_experiment: status and per-variant results.
• get_experiment_winner: the winning caption.

“A/B test three hooks for my launch post on X, launch them, and tell me which one wins.”

Recycle the winner

Keep a proven caption going out on a cadence.