The Optimizing for CPA section shows how the model redistributes spend to reduce overall cost per action (CPA) while meeting conversion and revenue targets. This view helps you understand the efficiency improvements that could result from reallocation across channels during a given period.
The objective of this scenario is to:
Achieve the highest number of conversions for the lowest possible CPA
Stay within a predefined or historical budget constraint
The model generates an optimal allocation and forecasts its impact across three metrics: conversions, revenue, and spend
Conversions view
This line chart compares the volume of conversions achieved on each day of the optimization period under three scenarios:
Optimal: Conversions projected by the model using the recommended spend allocation
Actual: Conversions achieved with the current budget
Reference: Conversions from the reference period, which is used as a baseline
The chart allows you to track performance over time. For example, on 15 February, the model projects 97,730 conversions, while the actual and reference values are both approximately 88,870, indicating a clear opportunity for improvement under the optimized plan.
Revenue view
This chart presents a similar time-based comparison for revenue. It helps determine whether the optimised plan would maintain or increase revenue while improving efficiency. For example, on 14 February, the optimal revenue is $3.86 million, and the actual and reference revenue is $3.51 million.
The higher revenue in the optimised scenario, combined with increased conversions, suggests better performance without added spend.
Spend view
The spend chart visualizes how the model distributes daily budgets across channels to achieve the lower CPA. The lines reflect spend across multiple sources, including TV, branded search, generic search, affiliate, display, and others.
This view enables you to:
See where the model recommends reducing spend, for example, in less efficient channels
Identify channels that receive increased allocation, for example, due to stronger conversion efficiency
Track how changes play out over time
Let’s take an example.
On 18 February, TV receives the largest share at $468,923. Other channels, such as branded search, generic search, and affiliate, receive much smaller but targeted allocations. Channels like SEO, email, and direct are excluded entirely, as they are either unmodelled or not part of the paid allocation.
Together, these three charts illustrate how the model achieves a lower CPA through more efficient daily and channel-level budget allocation. You can use this section to:
Validate that conversions and revenue stay within acceptable ranges
Identify how the spend shifts support improved performance
Explore which days or tactics show the greatest gap between actual and optimal outcomes
This section is especially useful when preparing recommendations for reallocation or when demonstrating the value of model-guided optimization to stakeholders.