The Decay Curves graph, found under the “Automation Performance” section in the Budget tab, provides a visual representation of how campaign spending has been related to performance. This tool offers insight into how campaigns are performing and helps guide budget allocation decisions for better results. The curve is fitted per campaign / ad set, depending on the level budget is set in the ad business manager.
Graph overview
The Graph displays:
X-axis: Spend (budget allocated to the campaign).
Y-axis: The goal of the automation, in this case, "Orders Completed."
Data Points: The observed values for the selected campaign, shown as colored dots.
Yellow Dots: represent results from the Last Click attribution model.
Green Dots: represent results from the MTA (Multi-Touch Attribution) model.
Estimated Mean and Decay Effect
The blue line labeled as the “Estimated Mean” represents the expected performance of the campaign, modeled using a logarithmic function. This curve follows the economic principle known as the Law of Diminishing Returns, which states that as you spend more, the additional performance (or returns) decrease over time. Essentially, for each additional euro spent, the model predicts a slightly higher return, but the incremental gain becomes smaller.
Surrounding the estimated mean are upper and lower bounds (lighter shades around the blue line), representing uncertainty in the prediction. This indicates the potential variability in performance outcomes. However, these do not mean the limits the model expects the results to fall into (you will find many of the measured results outside these bounds). The bounds represent the uncertainty of the Estimated Mean.
Recommended Budget
The vertical dotted line represents the recommended budget, a daily update based on the optimization model. This budget is calculated as the ideal amount to allocate for the campaign to maximize the total automation effectiveness. The model adjusts this recommendation each day, incorporating new results and accounting for the performance of all other campaigns within the automation.
Campaign Adjustments
As the decay curve is updated daily, it adapts to the actual campaign performance. If the measured results consistently exceed prior estimates, the curve will shift upward, leading to a higher recommended budget. Conversely, if a campaign underperforms and is expected to deliver diminishing returns, the recommended budget will decrease. These adjustments ensure that the system stays responsive to ongoing campaign trends.