Free tool · no signup · all tools
Two-sided verdict planner
What it does. A marketplace has two sides and each one needs its own floor. Enter what each side
produces in a month (name:per_month:cost_per_contact, one per line) and the planner returns the
contact floor per arm for the test you are running, how many months each side needs to see half of the floor in
one arm, the price of one full two-arm test, and which side can actually be read first.
The floor comes from the same formula as npx marketing-mindset sample: base 3%,
+20% relative lift, 80% power, alpha 0.05 → 13,914 contacts per arm, 27,828 in total.
Your sides
How to read the result
The floor is per arm, so each side has to deliver half of it: a side producing 400 units a month puts 200 in one arm, and 13,914 ÷ 200 is 70 months. Two sides do not make the test cheaper — they make it two tests, and if both draw on the scarce side, they split one arm's volume and neither finishes.
The planner deliberately does not invent a lift. If you do not have a lift to test, the honest answer is not a
number here: it is the smallest lift your volume can read, which the CLI reports with
npx marketing-mindset mde --base 0.03 --per-arm 5000.
Related
Marketing engineer for marketplaces · the page this tool supports · channel verdict fit · test cost calculator · readable unit chooser · test queue planner · the same maths as an npm CLI · Telegram