Open this lesson in your favourite AI. It'll walk you through the why, explain the demo, and quiz you on the try-it list.
The team that runs more experiments wins — not because every experiment succeeds, but because learning rate compounds. Experiment velocity is the number of rigorous tests a team ships per month; most teams running one or two per quarter will always lag a team running eight, even if both have equal insight quality. The bottleneck is usually process, not ideas.
Model the compound learning advantage of a high-velocity team over 6 months.
Use these three in order. Each builds on the one before.
In one paragraph, explain experiment velocity like I'm new to it.
Walk me through how to increase experiment velocity step by step — from ideation through instrumentation through decision.
Given a team blocked by a 3-week engineering cycle for every experiment, how would you redesign the process to reach 6 experiments per month without adding headcount?
function learningCompound(experimentsPerMonth: number, successRate: number, liftPerWin: number, months: number) {
let baseline = 1.0; // normalized conversion rate
let wins = 0;
for (let m = 0; m < months; m++) {
const monthlyWins = Math.floor(experimentsPerMonth * successRate);
wins += monthlyWins;
baseline *= Math.pow(1 + liftPerWin, monthlyWins);
}
return { totalWins: wins, cumulativeLift: ((baseline - 1) * 100).toFixed(1) + "%" };
}
const slowTeam = learningCompound(1, 0.3, 0.05, 6); // 1 exp/mo, 5% lift/win
const fastTeam = learningCompound(8, 0.3, 0.05, 6); // 8 exp/mo, same win rate
console.log("Slow team:", slowTeam); // { totalWins: 1, cumulativeLift: "5.0%" }
console.log("Fast team:", fastTeam); // { totalWins: 14, cumulativeLift: "97.9%" }