whatpower·for

Methodology

How we collect data, compute runtime, and score against reality.

Data collection

Every device entry pulls from four sources, in order of precedence:

  1. Manufacturer official spec sheet (latest PDF / product page).
  2. Amazon product listing (for price and any deviations from manufacturer spec).
  3. YouTube independent reviewer test data — primarily Hobotech, Will Prowse, Project Farm, and dedicated power station channels.
  4. Professional lab tests where available (TechRadar, OutdoorGearLab, Wirecutter).

Runtime calculation

Every product page lists scenario runtimes computed by:

runtime_hours = (capacity_wh × inverter_efficiency × usable_pct) / device_watts

  • inverter_efficiency: LiFePO4 = 0.88, NMC = 0.85
  • usable_pct: 0.92 (BMS reserve)
  • device_watts: scenario-specific (see scenario pages)

Spec Reality Score

For each tested device, we compute:

  • Capacity ratio: real_measured_wh / spec_wh
  • Continuous output ratio: real_sustained_w / spec_w
  • Surge ratio: real_surge_w / spec_surge_w

The Spec Reality Score is a weighted average:

score = 60 × capacity_ratio + 25 × continuous_ratio + 15 × surge_ratio

Scores are capped at 100. Devices that aren't yet third-party tested are marked "pending" and don't display a score until we have at least two independent test sources.

Solar pairing

Daily energy = panel_w × peak_sun_hours × 0.78 × 0.92. The 0.78 factor represents realistic panel output vs. nameplate, and 0.92 is the power station's charge controller efficiency. See the solar pairing guide for the breakdown.

Updates and changelog

Prices: daily. Specs: weekly. Real-world test data: monthly as new tests are published. Spec Reality Score is recalculated whenever new test data is added.