See how noisy weekly metrics can wander just by chance. Each step is +1 or −1 (coin flip), optionally with a small drift.
A random walk is a path formed by successive random steps. At each tick, a fair coin is flipped — heads means +1, tails means −1. Despite being purely random, the cumulative path can look surprisingly like a real trend, which is why random walks are a useful mental model for understanding noisy data.
Adding a small drift biases each step slightly up or down, simulating a weak underlying trend beneath the noise.