How we rate evidence and test
A product can never carry a stronger label than the evidence we hold. Here is what each label means and how a product earns it.
Evidence levels
| Label | Requires | Products today |
|---|---|---|
| Hands-on tested | We used the exact model, followed a published method, and recorded dated results and photos. | None |
| Independent lab report | We read a report from a named lab covering this exact product, with the sample, configuration and results. | None |
| No label (maker's published specs) | Figures come from the maker or a retailer, cited and dated. We have not confirmed them. | Most |
| Unverified | We could not find a reliable source for key details. | A few |
Methods we will publish before testing
We will not call anything tested until the method for that kind of product is written up here. Planned methods:
Plug-in kits
- AC output logged at the plug with a named meter across full days of real weather, with panel angle, orientation, location and date recorded. No yearly savings claim from a single day.
- Shutdown time when grid power is removed, measured with named equipment, before any claim about anti-islanding behavior.
Batteries and power stations
- Full discharge into a fixed AC load, recording delivered watt-hours against the rated capacity and the ambient temperature.
- A real fridge run overnight, with its label rating and the measured draw both recorded.
What we will never do
- Invent ratings, reviews, test results or lab names.
- Upgrade a label because a manufacturer's spec sheet is detailed.
- Publish a yearly savings figure from a single sunny day.