A marketing mix model turns weekly spend and sales into a return for each channel. Before it sees the data, you set a prior: your starting assumption about what a channel returns. Drag the prior for television below and watch what the model reports.
Illustrative model, figures invented to show the mechanicLeft, you assume TV barely pays back. Right, you assume it pays back handsomely.
A Webflow page cannot run a live model, so this is a small stand-in with invented numbers. The geo experiment here is a fixed illustrative reading of 2.1x. The lesson is real: when historical spend alone cannot pin the answer, your assumption decides it, and an experiment is what takes the deciding away.