Who asked: Leo Fischer, head of growth at Kestrel Software.
Why: the Team plan launches next month and the board wants the price backed by the test, not by competitors' price pages.
The data: price_tests: four prices shown at random to about 4,100 visitors each, and the purchases at each.
What done looks like: conversion and revenue per visitor per price, the best price, elasticity between the two leading prices, a test of whether conversion really drops, a yearly projection, and the launch price.
In a price test, visitors are shown one price at random, so differences in conversion can be put down to the price. The number that matters is rarely conversion alone: a higher price can convert fewer people and still earn more per visitor. Elasticity, the percent change in conversion divided by the percent change in price, says whether demand is sensitive at that point (below -1) or not (between -1 and 0). Each step adds to the same script.