Trade promotion optimization (TPO) is the use of data and predictive models to decide which trade promotions a CPG brand should run, at what depth and frequency, and with which retailers, so that promotional spend generates incremental revenue instead of subsidizing sales that would have happened anyway. It goes one step beyond trade promotion management (TPM): where TPM plans, executes, and tracks promotions, TPO tells you which promotions are actually worth running.
Trade promotion management is the system of record. It handles planning, budgeting, execution, and settlement, the workflow of getting a promotion live and paying for it. Trade promotion optimization is the decision layer on top. It asks a harder question: of all the promotions we could run, which ones will actually add revenue we would not have gotten otherwise? A brand can have excellent TPM and still waste money, because a well-executed promotion is still a loss if it only discounted sales that were already going to happen.
CPG brands spend an estimated $800 billion a year on trade promotions globally, and much of it is committed without a reliable read on incrementality. The core difficulty is the baseline: to know whether a promotion worked, you have to know what would have sold without it, and that counterfactual is invisible in a normal sales report. TPO exists to estimate that baseline, separate genuine lift from noise, and flag the promotions that quietly lose money every cycle.
TPO starts by modeling a baseline of expected sales for each product, store, and week. It then measures promotional lift against that baseline, adjusts for cannibalization, when a promoted item steals sales from the brand's own other products, and pantry-loading, when shoppers simply buy earlier, and calculates the true ROI of each event. The stronger platforms do this predictively, simulating a promotion's likely return before it runs, rather than only reporting on it afterward. That shift, from descriptive to predictive, is what separates real optimization from post-mortem reporting.
Two things, mostly. The first is store-level sales data, because promotions play out differently store by store and retailer by retailer, and national averages hide the decisions that matter. The second is predictive models rather than descriptive dashboards, since the point of TPO is to decide the next promotion, not to explain the last one. Without both, a brand ends up with a very detailed record of money it already spent.