Bayesian Statistics in Excel: Priors, Posteriors & Bayesian A/B Tests

What you'll build

What you will be able to do

Each lab ends with a concrete result that you build yourself and the checker confirms, cell by cell.

• Work out what a positive root-rot test really means for Juniper Lane's incoming plant batches: P(positive), P(rot | positive), P(healthy | negative) and the false alarms per year.
• Update the rot probability of one suspicious batch through three pieces of evidence, first step by step and then in one go with odds and likelihood ratios.
• Turn a team guess about the add-on purchase rate into a posterior on a grid of 10 rates, using BINOM.DIST as the likelihood, and read the posterior mean, mode and P(p ≥ 20%).
• Update a Beta prior for the 90-day reorder rate with launch data, and compute the posterior mean, standard deviation and a 95% credible interval.
• Update a flat and a strong prior with the same 40 visitors, compare the posterior means and P(p > 20%), and chart both posterior curves.
• Turn the checkout test (A: 120 of 2,500, B: 150 of 2,500) into two Beta posteriors and estimate P(B > A), once with the normal approximation and once by counting 30 posterior draws.
• Compute the expected loss of shipping B and of keeping A from the 30 posterior draws, the expected lift, and let an IF formula make the call against a threshold.
• Compute a frequentist confidence interval and one-sided p-value for the checkout test next to the Bayesian P(B > A), and pick the interval that may be read as "95% probability".