Sales and Brand Impact Score Calculation in Amplify TV
Sales Impact (SI) and Brand Impact (BI) scores are single metrics that predict advertising effectiveness by combining multiple performance measures into one weighted value. Each respondent receives an individual SI or BI score, which can then be aggregated and benchmarked
Sales Impact (SI) and Brand Impact (BI) are calculated in the same manner using this methodology.
Note: While both one number scores are calculated at the respondent level, they are most often shown as percentile scores calculated using a CDF percentile calculation against the relevant norm.
Calculation Process
Step 1: Measure Collection
- 5-Point Scales
- Binary Measures
- Frequency Measures
All raw respondent data collected across measure types
Step 2: Measure Normalization
All measures are normalized to the same scale to ensure comparability acros different question types and response formats.
Step 3: Coefficient application and score calculation
Each normalized measure X its SI or BI coefficient.
Amplify TV Coefficients
| Measure | Coefficient | |
|---|---|---|
| Watched Full Ad | 0.05 | |
| Unaided Brand Recall (tota) | 0.15 | |
| Overall Emotion | 0.25 | |
| Emotional Intensity | 0.05 | |
| Purchase Uplift | 0.15 | |
| Ad Distinctiveness | 0.20 | |
| Persuasion Appeal Composite | 0.15 |
Output: Respondent-level SI or BI score
The score is fixed to the respondent permanently
Step 4: Percentile Calculation
The project level mean of respondent's SI or BI scores is compared to a normative database of other Sl or BI scores using a Cumulative Distribution Function (CDF) percentile calculation.
The percentile score represents the ad's rank position within the distribution. A 66th percentile signifies that the area under the Cumulative Distribution Function (CDF) curve to the left of the ad's absolute SI score is 0.66. This means that the ad's score is greater than or equal to 66% of the absolute SI scores in the entire normative database.
Final Output: Percentile rank (0-100)
Key Technical Points
- Fixed calculation
- The SI score is calculated once per respondent and is never recalculated. Transformations and coefficients are constants derived from historical data.
- Transformation logic
- Binary and frequency measures are converted to 5-point equivalents by mapping their Z-scores against the standard deviation and mean of exsisting 5-point measures in the Amplify database
- Coefficient weights
- Each measure contributes a specific percentage to the total SI score. These weights were derived through data science validation against market mix modeling outcomes.
- Percentile interpretation
- The CDF percentile calculation provides an intuitive way to understand performance by showing where a stimulus ranks within the normative distribution of comparable tests.