PhoneCompare.PhonesCompare

How we score phones

Version 2026.09 · published September 8, 2026 · every score on this site is computed, never hand-written.

1. Spec scores — calibrated ranges

Each measurable spec is scored 0–100 against a fixed reference range that represents the realistic spread of modern phones. This means a score does not depend on the opponent: a great phone keeps its score in every matchup, and the same device gets the same score on its own page as in any comparison.

SpecRange (0 → 100)Direction
Performance score500,0002,500,000 ptshigher is better
RAM416 GBhigher is better
Storage64512 GBhigher is better
Screen size5.57.2 inhigher is better
Pixel density320560 ppihigher is better
Refresh rate60144 Hzhigher is better
Brightness6005,000 nitshigher is better
Selfie camera850 MPhigher is better
Video quality60100 ptshigher is better
Camera quality80160 ptshigher is better
Battery3,0006,500 mAhhigher is better
Charging speed15150 Whigher is better
Price2501,500 USDlower is better

2. Category scores — weighted averages

Spec scores are combined into 5 categories. Each category has a weight that reflects how much it matters in everyday use:

  • Performance
    30%
  • Display
    20%
  • Camera
    20%
  • Battery
    20%
  • Price
    10%

Camera quality is judged by an independent testing score — not megapixels. Chipset speed outweighs RAM. Informational rows (chipset names, megapixel counts) are displayed but never decide a winner.

3. The grading curve

No real phone tops every range (price pulls flagships down, charging speed pulls everyone else down), so raw averages would leave even the best device near 75. We apply one final curve so the best phone in the database maps to 95/100 — like school grades. The curve never changes the order of devices, only the scale.

4. Verdicts

In a comparison, a device must lead a category by at least 2 points to claim it. The overall winner is the device with the most category wins — or the clearly higher total (≥3 points). Otherwise we say “too close to call” and tell you which category to decide on.

5. Data & updates

Specs come from manufacturer pages and are re-verified at least monthly; every page shows its last-updated date. Scores are machine-computed from that data and versioned (2026.09) — when the methodology changes, the version changes and all pages are regenerated, so a cited score is always traceable to a published version. Machine-readable scores are available as JSON (see llms.txt).

Limitations

  • Prices are typical launch MSRPs — actual street prices vary by market and time.
  • Benchmark scores are approximate composites; real-world performance varies by workload.
  • Software experience, update support, resale value and brand ecosystem are not scored yet.
  • Specs are demo-grade during early access — always verify with the manufacturer before buying.