| Priority | Source | Used for |
|---|---|---|
| 1 | Manufacturer specification pages (ecoflow.com, ankersolix.com, jackery.com, renogy.com, etc.) | Capacity, output, surge, chemistry, cycle ratings, weight |
| 2 | Certification databases (UL Product iQ, Intertek listings, FCC ID) | Safety listings — see the certification decoder |
| 3 | Government guidance (CPSC, CDC, USDA FSIS, FDA, NHC, Ready.gov) | Every safety, food-safety, medical-storage and hurricane-timing claim |
| 4 | Live retail listings, checked by hand | Price tiers, availability, product identity |
| 5 | Verified-purchase reviews, read for failure patterns | The cons lists — recurring complaints, not one-off anecdotes |
How we evaluate backup power gear
We do not lab-test these products, and we will never imply that we do.What we do instead: build a verified spec corpus from primary sources, run every runtime number through one disclosed formula, verify each recommendation against the live retail listing, and state plainly what we cannot know. Every assumption we make is on this page.
Methodology version 1.0 · Published · Maintained by Hilario Herrera. Changes to this methodology are logged at the bottom of the page.
1. Where our numbers come from
Every specification on this site traces to one of these sources, in priority order:
When a manufacturer's page and its retail listing disagree, we use the manufacturer figure and note the conflict. When a claim exists only in marketing copy with no spec behind it, we leave it out.
2. The runtime formula, and every assumption in it
Every runtime figure on this site comes from one formula:
Runtime (hours) = (rated Wh × efficiency factor) ÷ average load watts
These are the exact values we plug in, and why:
- AC inverter efficiency: 0.85. Modern portable power station inverters run 80–90% efficient under load; we use the conservative middle.
- DC/USB efficiency: 0.95. DC ports bypass the inverter; losses are small but not zero.
- Refrigerator average draw: ~50Wh per hour — 150W running at a 33% compressor duty cycle. Real duty cycles run 30–50%, higher in a hot Florida interior, so we present runtimes as ranges.
- Appliance wattages are input draw, not marketing ratings. A "1,000W" microwave draws roughly 1,500W; our calculator uses the input figure.
- Startup surge: roughly 3–6× running watts for compressor appliances, checked against each station's published peak rating before we say it can run the load at all.
- Solar harvest: 4.5–5.5 peak sun hours × a 70–80% real-world derate. A 200W panel is credited 700–900Wh per clear Florida day, and 10–25% of rated under heavy overcast.
3. How products get picked (and dropped)
- Chemistry first: we favor LiFePO4 for its thermal stability and 3,000+ cycle ratings.
- Certification: UL 2743 or an ETL listing to it weighs heavily — see what the marks mean.
- Availability during hurricane season: a perfect unit that sells out every June is a bad recommendation here.
- Value: we check cost per watt-hour against the observed street price when we rank a station, and the result shows up as the ballpark tier and the ranking on the comparison page — not as a number we would have to keep current.
- Every product carries at least one true limitation. If we cannot name a real con, we have not researched it enough to recommend it.
- Monthly re-verification: every linked listing is re-checked monthly for availability, identity and price. Dead or changed listings get replaced or dropped — the check date appears on the picks page and in the open dataset.
4. Price tiers, not fake live prices
Every pick carries a ballpark tier instead of a number. $ is the budgettier — the safety-and-communications layer plus the smallest batteries, enough for phones, a router, a fan and one CPAP night. $$ is mid-range — the 1kWh and 2kWh class, dual-fuel inverter generators and expansion batteries, the tier that keeps a refrigerator alive. $$$ is premium — the largest stations and name-brand generators, bought for medical dependence, well pumps or week-long restorations.
We show tiers because we cannot see Amazon's price in real time, Amazon's own rules say not to publish prices we cannot keep current, and a stale number is worse than an honest tier. The tier is set from the listing on the check date shown; Amazon's price at checkout is the price. We keep the underlying dollar band internally, unpublished, purely so we can tell when a product has drifted out of its tier and needs re-checking.
5. What we deliberately do not claim
- We have not run these units in a lab, measured their inverter curves, or cycled their batteries.
- We cannot verify manufacturer cycle-life ratings; we report them as ratings, not findings.
- Product images on this site are labeled illustrative renders, not photographs of the units — retailer image licensing does not yet permit us to show real product photos.
- Runtime figures are estimates from the formula above, not measurements of your appliance. Your label beats our math.
- Medical and tax content is general information — we are not clinicians or tax professionals, and we say so where it matters.
6. How the site makes money, and what that buys
SunGrid Living earns commissions on qualifying Amazon purchases, disclosed on every page that carries a paid link (full disclosure). No manufacturer pays for placement, supplies units, or sees content before publication. The commission is the same to us regardless of which product you choose, which is what lets the cons lists stay honest.
7. Corrections
Found a number that is wrong? Email [email protected]. Verified errors get corrected in place, and material corrections are noted on the affected page. In August 2026 we ran a full adversarial audit of every numeric claim on the site against primary sources and corrected everything it caught — that standard is now the bar for new content.
Methodology changelog
- Aug 19, 2026 — v1.0. First published version, formalizing the formula, sources and assumptions used since launch, following a sitewide audit of all numeric claims (Aug 15, 2026).