The idea
Nomic is a concept for a closed-loop Bayesian decision layer for MVNO and FMCG commercial operations—turning customer, usage, billing, campaign and competitor signals into governed portfolio moves and targeted treatments.
Nomic estimates customer state, churn risk and action effects from observable evidence. It combines posterior beliefs, Monte Carlo planning, margin economics and exploration bonuses to fund a mixed plan of pricing moves and customer treatments—then launches with control holdouts and learns from measured outcomes.
Estimate honestly
Posterior beliefs expose uncertainty, evidence counts, credible intervals and blind spots.
Optimise value
Scenarios compare contribution margin, downside, budget cost and expected leakage.
Close the loop
Guardrails, approvals, holdouts, append-only audit and validity warnings prevent false learning.
Example scenario
New customer and market observations update prior beliefs. A retention offer is evaluated against margin and downside, with a control holdout to measure the outcome.