SoundScore
Overview
For each device, we reduce the full set of five component metrics for two fits down to a single number -- the SoundScore. We take a three-step approach:
- Compute a single per-fit score (weighted average of component metrics)
- Combine those scores across fits to make the device's overall score
- Add a fixed +1.1 presentation offset to place the final scores into a familiar range
Step 1: Per-Fit Score (Component Weights)
The per-fit score is a weighted average across all 5 component metrics. The weights were informed by a custom survey of hearing aid consumers (n = 107) and hearing care professionals (n = 95). Each participant ranked the relative importance of our 5 metrics on a 0 to 4 scale (higher is better).
Table 2. Rated importance of each component dimension and scoring weights. Group average rated importance from 0 (lowest) to 4 (highest). The bottom row is the two-group average normalized to sum to 1.0.
| Speech Benefit (Quiet/Mod) | Speech Benefit (Noisy) | Own Voice Not Boomy | Does Not Squeal | Streaming Music Quality | |
|---|---|---|---|---|---|
| Consumers Avg Rank | 3.2 | 3.1 | 1.1 | 1.3 | 1.3 |
| HCPs Avg Rank | 3.5 | 2.7 | 1.4 | 1.7 | 0.7 |
| Combined Normalized Weight | 0.34 | 0.29 | 0.12 | 0.15 | 0.10 |
The ranks show reasonably nice agreement across groups. We averaged the two groups and normalized the result to produce the scoring weights: own-voice naturalness receives 0.12 and feedback handling receives 0.15.
The per-fit score is:
Step 2: Across-Fit Score (Initial vs. Tuned Weights)
The across-fit score is a weighted average of the Initial Fit and Tuned Fit per-fit scores. These weights were derived from a separate survey.
257 hearing aid consumers were asked to rate: "How Important Is It That Your Hearing Aids Sound Good With Minimal Effort?" on a 5-point scale. We intended this question to be a proxy for the importance of the quality of the initial fit.
Table 3. Percentage of hearing aid consumers responding to: "How Important Is It That Your Hearing Aids Sound Good With Minimal Effort?"
| Not-At-All Important (0.0) | Slightly Important (0.25) | Moderately Important (0.5) | Very Important (0.75) | Extremely Important (1.0) |
|---|---|---|---|---|
| 1% | 4% | 15% | 37% | 43% |
When responses were normalized to a 0--1 range, the group average is 0.77. Therefore:
| Fit Condition | Weight |
|---|---|
| Initial Fit | 0.77 |
| Tuned Fit | 0.23 |
Step 3: Normalization
When this methodology was first published, the observed overall scores had a maximum of 3.9 and an average of 2.3. These values could be misleading to users because in other products, the star rating of the best performing devices is usually very close to 5.0 (a score of 3.9 might be seen as mediocre).
We add a fixed +1.1 presentation offset to every device's raw across-fit score:
The same +1.1 is added to each Initial and Tuned per-fit score. This adjustment preserves product rankings and the differences between scores.
The offset remains fixed when products are added, published, unpublished, or retested. It is independent of the current leader. Changes to another product's measurements cannot shift an existing device's SoundScore. Results are not capped at 5.0.
This normalization is a presentation choice to align with consumer expectations for star-rating scales, not a scientific transformation. The underlying weighted scores are preserved in the data.
Expert Choice Award
We give an Expert Choice award to any device with a displayed SoundScore 4.0. This is a fixed threshold.
The calculated SoundScore is first rounded to one decimal using half-up rounding, and the threshold is applied to that public score. For example, a calculated score of 3.95 rounds to 4.0, receives an A SoundGrade, and qualifies for Expert Choice.