Once a media mix model is optimized, the following metrics are generated in the Optimization Detail section:
Status: indicates whether the optimization ran successfully.
Time Periods: the number of periods included in the optimization window (default: 13).
Total budget: the total budget used in the optimization scenario.
Incr. Outcome/Period: the predicted KPI lift per optimization period compared to the baseline (historical allocation). In other words, this shows how much additional KPI the optimized budget allocation is expected to generate per period relative to your actual historical spend levels.
Incr. Outcome %/Period: the predicted percentage lift in KPI per optimization period compared to the baseline. This indicates the improvement that the optimized allocation is expected to deliver compared to your historical allocation.
For models built using the lightweightmmm library (Bayesian regression), additional metrics shown are:
Iterations: the number of steps the optimization algorithm performed while searching for the best budget allocation. Each iteration represents one adjustment of the budget distribution as the algorithm moves toward a better solution. A higher number of iterations typically means the algorithm explored more potential allocations before converging.
Inferences: the number of times the objective function was evaluated during optimization. This reflects the number of candidate budget allocations that were tested to determine their predicted KPI impact.
Interpreting your Optimization Detail section:
The table compares your baseline allocation with the optimized allocation for each channel. Each row represents either the KPI or a media channel.
Below is a breakdown of each column:
Baseline: baseline shows the average per-period KPI and media spend at the actual spend levels. For channels, this represents the average spend per period during the selected timeframe. For the KPI row, this represents the average KPI per period under the baseline (actual) allocation.
Mix: this shows the percentage share of the total budget allocated to each channel under the baseline allocation.
Optimized: the optimized column displays the average KPI and media spend per period under the model’s recommended allocation. For channels, this means the new suggested average spend per period. For the KPI row, this means the predicted average KPI per period after reallocation.
Mix: this shows the percentage share of the total budget allocated to each channel under the optimized scenario. Comparing this column with the baseline Mix column helps you understand how the budget distribution changes.
Var: the difference between the optimized and baseline values. For channels: the change in average spend per period. For the KPI: the change in average KPI per period.
Var %: the percentage change between baseline and optimized values. This helps you understand the relative magnitude of the shift.
Mix Shift: Mix Shift highlights how budget allocation changes across channels between the baseline and optimized scenarios. A positive shift means that the channel receives a larger share of the total budget. A negative shift means that the channel receives a smaller share.