Methodology
How we score mattresses
Every number on this site comes from the same rubric, version 2.1. We publish the weighting and the raw measurements behind each category so you can check our math instead of trusting a badge.
What each category is worth
| Category | Weight | What we measure |
|---|---|---|
| Comfort / pressure relief | 22% | Peak shoulder pressure (mmHg) on an XSensor pad, side-sleeping 165 lb subject |
| Support / spinal alignment | 20% | Lumbar sinkage (in) under a 230 lb back-sleeping subject, plus alignment photography |
| Cooling | 15% | Surface temperature rise (°F) after 60 minutes of body heat in a 68°F room |
| Motion isolation | 12% | Peak g-force at 12 in from a 10 lb dropped weight |
| Edge support | 11% | Perimeter compression (in) under a 230 lb seated load |
| Durability | 10% | 30,000-cycle roller test simulating eight years of use |
| Value | 10% | Score-per-dollar against the median queen price in the same category |
What the numbers mean
- 9.0 – 10
- Best in class. Top decile on the measured metric with no meaningful trade-off.
- 8.0 – 8.9
- Strong. Beats the category median clearly; a safe recommendation for the right sleeper.
- 7.0 – 7.9
- Competent. Around the category median — fine, but something else does it better.
- 6.0 – 6.9
- Below average. A real weakness for the sleepers who care about that metric.
- Below 6.0
- A failure point. We will name it in the cons and steer most readers elsewhere.
Rules we hold ourselves to
- A score is never assigned before the mattress has been through the full protocol. If a model has not been measured yet, its review says so rather than showing an estimate.
- Category scores are position-agnostic; where a mattress suits one sleeper and not another, that goes in the verdict, not into a fudged number.
- Brands cannot see or influence scores before publication, and no score has ever been changed at a brand's request.
- Scores are re-checked when a manufacturer changes construction. The review's last-updated date reflects the most recent re-check.