Amazon’s DSP audience segmentation is quite complex, requiring a comprehensive consideration of multiple metrics for audience selection and optimization. However, due to the decision-making factors being scattered across different metrics, making quick optimization decisions is not easy.
The quadrant analysis method, also known as matrix analysis, uses two key attributes of an entity as dimensions for classification and analysis. There are generally two scenarios:
1) Classification: Dividing businesses or products based on designated metric performance;
2) Diagnosis: Identifying business issues through changing business metrics. This is often visualized using bubbles or scatter plots on the XY axis, and then by quadrant division, to pinpoint the strengths and issues within each quadrant.
The well-known “BCG Matrix” is a type of quadrant analysis.
The BCG Matrix describes all of a company’s products in terms of “market growth rate” and “market share,” subsequently categorizing them as “Stars,” “Question Marks,” “Cash Cows,” and “Dogs.” The BCG Matrix assists companies in evaluating the development status of different products, adopting various strategies, and optimizing resource allocation.
We’ve applied the BCG Matrix to DSP audience segmentation:
Step 1 – We chose a bubble chart for visual representation, with different audiences displayed by distinct bubbles. The key metric, click-through rate, serves as the X-axis, while cost per click is on the Y-axis. The volume of purchases determines the size of the bubble, and different bubble colors represent different audience segments.
Step 2 – The average click-through rate and average cost per click are used as the central axes for the X and Y axes, respectively, dividing the entire view into four quadrants.
- First Quadrant: High click-through rate, high cost per click
- Second Quadrant: Low click-through rate, high cost per click
- Third Quadrant: Low click-through rate, low cost per click
- Fourth Quadrant: High click-through rate, low cost per click