How we test and rank

Updated Oct 2, 2026

In short: we don't test units ourselves yet. We collect capacity tests published by independent labs and reviewers, keep the ones that are comparable, and use the median. Every figure on this site links to the test it came from.

Why rated capacity isn't what you get

The rated capacity on the box (in watt-hours, Wh) is the energy the battery cells hold. Getting it out of a wall-style outlet means converting battery power to 120 V AC, and the inverter, screen, fans and control board all use some of it along the way. In the tests we track, units delivered 75% to 96% of their rated capacity, with a median of 86%.

Where the figures come from

Our main sources are CNET Labs, OutdoorGearLab and Outdoor Life, with extra results from StorageReview, Hobotech, PCWorld and others. CNET runs every unit at the same fixed 480 W load until it is empty, which makes its results the easiest to compare. We record only the numbers, the test date, the load and a link. The testing, the write-up and any photos belong to the tester.

We read every result by hand. We don't scrape testers' sites.

Which tests count

• US units at 120 V. UK and European units run at 230 V and can behave differently. We show them as a cross-check but leave them out of the figure.
• AC output. Most people use the AC outlets, and that's where the losses are.
• Loads of 100 W or more. At light loads the unit's own running costs take a much bigger share. A BLUETTI AC180 gave 54% at 34 W but 84% to 89% at 200 W to 800 W. We list light-load results separately, because they matter for CPAP machines and overnight use.
• A described method. We leave out results where the tester doesn't say how they measured.

How we work out the numbers

• Delivered (%) is the median of the counted tests, each as a percentage of the rated capacity.
• Usable capacity (Wh) is the rated capacity times that median.
• Cost per usable Wh is the current price on the brand's US store divided by the usable capacity. It's the fairest way we know to compare value across sizes.

Many models rest on one test for now. We show how many tests sit behind every figure, and we add more as testers publish them.

Specs

We take specs from each maker's US user manual where we can, and the product page otherwise. Each value on a product page links to where we found it. Makers' pages are wrong more often than you'd expect: we've found 120 V outlets labeled as a 240 V plug type, a unit marketed as expandable that its own support page says isn't, and an output rating the US model can't deliver. Where sources disagree, we use the manual and say so.

Prices

For now we check prices by hand on each brand's US store and show the date we checked. Prices change often, so treat them as a guide and confirm on the store. We don't show Amazon prices, because Amazon only allows them through its own live price feed.

Price verdicts

Every priced model gets a plain verdict based on what each usable watt-hour costs at the price we checked: Great value at 50¢ or less, Good value up to 60¢, Fair price up to 75¢, and Pricey above that. The cut-offs are our own, set from the spread of prices across the models we track. When a price changes, the verdict changes with it.

Our score

Each review has an overall score out of 10. It comes from data, not opinion: the tested usable capacity and the specs from the maker's US manual, compared with the other models we track. Price isn't part of it, so the score doesn't move when prices do. Value has its own verdict, above.

• Each measure becomes a score from 5 to 10 based on where the model ranks among the models that have that figure. 7.5 is the middle of the group, and a model that beats nearly all of them gets close to 10. Ties share a rank. The score ranks models against each other, so a lower score means fewer strengths than its rivals, not a bad unit.
• Capacity, power and charging are compared within each size class, so a small unit isn't marked down for being small. Battery, build and features are compared across all models.
• Yes or no features score 10 if the model has them and 5 if it doesn't. A LiFePO4 battery scores 10 and NMC 7, because LiFePO4 lasts several times as many charge cycles.
• Missing figures are left out and the rest of that group counts for more. If more than 40% of a group is missing, the model gets no overall score.

The four groups and what's in them:

• Capacity and power (35%): tested usable capacity (60%), continuous AC output (25%), AC surge (15%).
• Charging (25%): wall charge to full (40%), max solar input (35%), UPS switchover (25%).
• Battery and build (25%): battery type (30%), cycle life to 80% (25%), warranty (25%), usable Wh per lb (20%).
• Features (15%): expandable battery (25%), AC + USB-C outlets (20%), USB-C max output (15%), 240 V output (15%), app control (15%), wheels (10%).

Shorter wall charge times and UPS switchover times score higher, and a unit with no UPS mode scores 5 on that line. The overall score is the weighted average of the four groups.

Many models rest on a single capacity test, so treat a difference of a few tenths as a tie. This is version 1 of the formula, from Oct 14, 2026. If we change it, we'll say what changed and when, here.

How we make money

We may earn a commission when you buy through links on this site. That never changes the numbers or the order of any list. Full disclosure.

Corrections

If you spot a wrong number, tell us and we'll check it against the source. When we correct a figure, we change the "Updated" date on the page.