Flat editorial diagram — review method overview (labeled infographic, not a product photo).

Owner threads, dated spec sheets, and the job

A 2 kWh box that can’t start your fridge is a paperweight, however good it looks in the render. This page is the rubric behind every money page. Read it once and you’ll know exactly what our rankings can and cannot promise you.

The short version

  • Rankings start from the job, then the manufacturers' own US product pages with the check date printed, plus the failure modes owners keep repeating: surge, open-circuit voltage, transfer-switch wiring.
  • A cell stays empty rather than guessed.
  • Recommendations are never sold, and no manufacturer sees a page before it goes up.

Affiliate disclosure: Example /go/ hops later on this page are the same affiliate exits used on money pages. If you buy after clicking, we may earn a commission at no extra cost to you. Full disclosure.

What a ranking here is made of

Let’s be plain, because this is the page where sites usually start lying. The work is the slower kind: manufacturer electrical ratings read closely, the failure modes buyers actually report (surge, open-circuit voltage, transfer-switch wiring), and the hard limits of each category. Street prices are copied from the US product page on a dated check, currently August 26, 2026, and they move with sales. A crossed-out compare-at price is what the manufacturer showed that day. Not a promise it comes back.

Every named Jackery, BLUETTI, and Anker SOLIX SKU carries the official product photograph from the manufacturer page. The lifestyle photos (the garage, the basement, the kitchen, the driveway) are mood, and they never stand in for a product shot. If we can't get a clean official image, we don't feature the SKU, full stop. That's why you won't find a named HomePower, AC500, Explorer 1000, or F2000 card here, and it's a trade we'd make again.

Ranking factors

  1. The job. Fridge + network, medical devices, a 240 V well pump, and whole-home HVAC are four different purchases. Runtime math for common loads lives on the fridge runtime and sump pump runtime pages; size the tank with the load calculator. A page that crowns one “best solar generator” for all four is selling a trophy, not a machine.
  2. Continuous watts vs surge. Compressors and well pumps care about the first second. When a manufacturer page lists conflicting surge numbers, we leave the cell blank instead of picking the larger one.
  3. Usable watt-hours. Nameplate capacity is not the same as a 90% discharge on LiFePO4. We still quote nameplate, because that's what the cart shows, but the runtime estimates in our copy stay conservative.
  4. Expandability and 120/240 V. A station that stays 120 V cannot feed a split-phase dryer, whatever the marketing says. Extra battery packs count only when they're a real SKU, not a coming-soon slide — see extra batteries.
  5. Recharge paths. Wall AC, your alternator, and portable panels with a Voc the station will actually accept. Series-wiring a pile of mismatched blankets is how charge controllers die.
  6. Support and warranty. A four-figure brick needs parts and firmware years from now, not just a glossy app today. We prefer brands with a US product page and a phone number somebody actually answers.
  7. Program quality. We feature 30-day-cookie brands (Jackery, BLUETTI, Anker SOLIX).
  8. Photography and sourcing. Official product photograph, specs and street prices dated, and unknown cells left blank. A blank cell in our tables is a finding, not a gap.

Caveats we will not sand off

Portable stations are not silent whole-home plants. Standby generators need permits, pads, and oil changes — start with the standby generators we'd shortlist or the standby vs portable fork if you are still choosing a machine class. “Home batteries” on this site are mostly expandable power-station ecosystems, not Tesla Powerwall. Rooftop solar is out of scope. Comparison CTAs go only to /go/jackery, /go/bluetti, and /go/anker-solix.

What we do, and what we do not

  • Spec-based unless stated otherwise. If we did not handle the unit, the money page says so in the first screen and in the honesty block. We do not write “we tested” without a photo of the unit in our hands.
  • Photos. Product cards use manufacturer press/studio images. Lifestyle leads are real photos we own or license. No AI-generated product renders. Quarantined FAIL photos stay quarantined — they do not return to the build.
  • Affiliates. When a /go/ hop is live we may earn a commission at no extra cost to you. Disclosure sits above the first buy-intent link, not only in the footer. Dormant programs stay dormant in copy.
  • Re-verify. “Last verified” is the real date the page was rebuilt. We do not bump dates on pages that did not materially change.
  • QC. Money pages are expected to pass an Astra gpt-6-astra QC row before a Donjulio batch. Open FAIL rows block DONE.

If a money page disagrees with this policy page, the money page loses and gets fixed. Spam-window shield: this page must match what the ten indexed URLs actually claim.

Who a page is for

Every money page carries a who-it's-for and who-should-skip pair, because the kindest thing a review can do is tell you to leave. When a $4,000 station is the wrong tool for your house, we say so and point you at whole-house generators or a smaller power station instead. If we got a number wrong, tell us at contact. Corrections are part of the method, not an embarrassment to it. The money side lives at the affiliate disclosure.