How Scoring Works
A concise, formula-level breakdown of every scoring dimension. No guesswork — if you want to know exactly why one laptop beats another, this is the page.
The short version
Six dimension scores (Performance, Display, Battery, Portability, Value, Build) are each computed from raw spec data on a 0–100 scale. A weighted sum produces the composite score. The weights depend on the active Use Case profile — a gaming laptop is scored differently than the same machine under an ultrabook profile.
Example: Gaming Profile
Using the actual gaming weight profile (weights sum to 100%):
| Dimension | Score | Weight | Contribution |
|---|---|---|---|
| Performance | 95 | 40% | 38.0 |
| Display | 82 | 20% | 16.4 |
| Battery | 48 | 8% | 3.8 |
| Portability | 60 | 7% | 4.2 |
| Value | 42 | 15% | 6.3 |
| Build | 80 | 10% | 8.0 |
| Composite | — | 100% | 76.7 |
* Example scores are illustrative. Use the "Why?" button on any card for the exact breakdown.
Dimension Formulas
* Weights shown are from the Gaming profile. Other profiles weight dimensions differently — see the Methodology page for all profiles.
Performance
40% weight *Blends CPU and GPU benchmark indexes. The blend ratio is profile-specific. Gaming is GPU-heavy (35% CPU / 65% GPU). Business is CPU-heavy (65% CPU / 35% GPU).
// Gaming profile blend
Performance = (CPU_index × 0.35) + (GPU_index × 0.65)
RTX 5090 GPU_index = 100, Core Ultra 9 CPU_index = 84 → Performance ≈ 94
Display
20% weight *Panel technology alone accounts for 40% of this score — the OLED vs IPS gap is significant. Refresh rate, brightness, color gamut, and resolution contribute the rest.
// Component weights
Display = panel(40%) + refresh(20%) + brightness(15%) + color_gamut(15%) + resolution(10%)
// Panel scores: OLED = 95, Mini-LED = 80, IPS = 60, TN = 25
// Refresh: 60Hz = 30 floor, 240Hz = 100 ceiling (linear)
OLED 240Hz 100% P3 2560×1600 → Display ≈ 91
Battery
8% weight *Primarily real-world battery life (75% weight), with raw capacity as a supporting signal. Floor is 3 hours; ceiling is 20 hours.
Battery = normalize(life_hours, 3h, 20h) × 0.75
+ normalize(capacity_Wh, 30, 100) × 0.25
18h battery / 66.5Wh (MacBook Air M4) → Battery ≈ 89
Portability
7% weight *Inverse of weight and thickness — heavier and thicker laptops score lower. Weight dominates at 70% since it's the primary portability pain point.
// Both are inverted — lighter = higher score
Portability = normalize(3.5 − weight_kg, 0, 2.7) × 0.70
+ normalize(35 − thickness_mm, 0, 25) × 0.30
Surface Pro 11 at 0.895kg, 9.3mm → Portability ≈ 96
Value
15% weight *Performance per dollar. Dataset-normalized so a budget laptop with good specs genuinely outranks an expensive flagship on this dimension. Uses MSRP for static scoring.
// Raw ratio
value_raw = performance_score / (msrp_usd / 1000)
// Normalize across full dataset
value_score = (raw − dataset_min) / (dataset_max − dataset_min) × 100
$499 IdeaPad Slim 3i → value_raw ≈ 28. $3,779 ROG G16 → value_raw ≈ 25
Build
10% weight *Chassis material + MIL-SPEC certification. MIL-STD-810 adds 15 bonus points — real-world durability has tangible value, especially in business and 2-in-1 segments.
// Material base scores
carbon-fiber = 90 magnesium = 85 aluminum = 80
mixed = 65 plastic = 50
Build = material_score + (mil_spec ? 15 : 0) // capped at 100
Carbon fiber + MIL-SPEC (ThinkPad X1) → Build = 90 + 15 = 100
Frequently Asked Questions
Can a laptop's score change without any spec changes?
Yes — the Value score is dataset-relative. If new laptops are added with better or worse value ratios, existing scores shift slightly to maintain proper normalization across the full set.
Do review scores affect the composite score?
Review scores from publications are displayed for reference but are not currently factored into the composite score. The composite is derived entirely from structured spec data.
Why does a gaming laptop score low on an Ultrabook profile?
The Ultrabook profile weights battery (25%) and portability (20%) heavily. A 2.5kg gaming laptop with 6h battery will score poorly on those dimensions, dragging the composite down even if its performance score is excellent.
How is the "Performance / $" sort different from the Value score?
The Value score uses a normalized 0–100 scale across the dataset. The "Performance / $" sort uses the raw ratio of (CPU + GPU index) / price — simpler and more linear. Both tell a similar story but with different sensitivities.
Are affiliate relationships factored into scores or rankings?
No. Scores are computed mechanically from spec data. Rankings cannot be purchased or influenced. See our editorial policy for full details.