Methodology
How we rank things
The rubric with its weights, the source for every class of number, and the things we refuse to claim. Specific enough that you could reproduce the ranking yourself.
Last updated August 31, 2026
We rank on published, checkable specifications and nothing else. The weights are below, the sources for every class of number are below, and so is the list of things we refuse to claim. We have tested zero units and no part of this method depends on pretending otherwise.
What we measure, and where it comes from
| Claim | Source we accept | Source we refuse |
|---|---|---|
| Rated wattage | The manufacturer's own listing or manual, cited per model | Category averages, other affiliate sites, any estimate |
| Cost per night, month, season | Our own arithmetic from the above, at the EIA rate, working shown | Any figure without its rate and its arithmetic |
| Electricity rate | US EIA published average residential price, with the retrieval date | UK or EU figures; unstated rates |
| Temperature range | The manufacturer's published range for that model | Anything inferred from wattage |
| Certification | The mark or standard the maker names on the listing | Assuming certification from brand reputation |
| Warranty, heat settings, timers, dimensions | The manufacturer's listing | Retailer summaries that conflict with the maker |
| Price and availability | The live retailer data layer only, timestamped | Any remembered or hard-coded number |
This table governs every page on the site, not just roundups.
The ranking rubric, with weights
We publish weights because a ranking whose criteria are secret is an opinion wearing a lab coat. These are the axes we can actually substantiate from published data.
| Axis | Weight | How it is scored |
|---|---|---|
| Published wattage | 35% | Does the maker state one at all, and how does it compare within the category |
| Control appropriateness | 25% | One controller or two, matched to the format; number and granularity of heat steps |
| Safety disclosure | 20% | Named certification and standard, auto-shutoff range, stated overheat protection |
| Stated warranty | 10% | A published term. No term stated scores zero rather than being assumed |
| Documented specifics | 10% | Temperature range, fabric weight, cord length, pocket depth: specificity is rewarded |
Weights are applied within a category, never across formats. A throw is not compared against a mattress pad.
Notice what is absent: warmth, softness, heat-up time, build quality and comfort. Those are the axes every competitor ranks on and we cannot assess a single one of them, because we have not handled these products.
What we did not do
Stated plainly, because it is the whole basis on which the rest should be read.
We have not handled these products. We claim to have tested zero units. Everyone in this category says they tested twenty blankets. We have not tested any, and we say so.
Here is what we did instead. We read the live listing for every product on this site, pulled the rated wattage where the manufacturer publishes one, recorded by name the ones that publish nothing, took the electricity rate from a government source with its retrieval date, and did the arithmetic in the open so that any reader can re-run it.
Why the rated figure, and what is wrong with it
An honest method has to state the limits of its own numbers, so: rated wattage is a ceiling, not an average. A thermostatic blanket cycles its element off once the bed is warm, so actual overnight consumption lands below our arithmetic, often well below.
We publish the rated figure regardless, for two reasons. It is the only number the manufacturer stands behind, and it bounds your bill from above: your real cost cannot exceed it. Estimating a duty cycle for a blanket we have not measured would be exactly the invented number this whole method exists to avoid.
How to check us
- 1Follow any wattage citation. Every figure links to the manufacturer listing it came from. If the listing has changed since we checked it, tell us.
- 2Re-run the arithmetic. Watts divided by a thousand, times hours, times the rate. The rate is linked under every table. Use the rate off your own bill and the answer becomes exactly right for you.
- 3Check the electricity rate at source. We use 0.1834 USD per kWh: US Energy Information Administration, Electric Power Monthly, Table 5.3 (June 2026): 18.34 cents per kWh, average US residential, retrieved 2026-08-31. It is linked below.
- 4Measure a product yourself. A plug-in energy monitor costs about $20 and reads actual draw. If your figure differs from ours, yours is better, because it is a measurement.
When we exclude a product
- We cannot resolve it to a real, currently-sold listing. Every product here was read off a live retailer page.
- It has an open safety recall. We check the CPSC database rather than maintaining our own list.
- Its only distinguishing claim is unquantified. A product whose entire pitch is an unmeasured low-EMF claim does not get ranked on that claim.
- The buy link cannot exist. We do not write a comparison whose recommendation you cannot act on.
Corrections and cadence
Report an error and we will check the source you name, fix the page, and date the fix. If we got it wrong we say so on the page rather than editing silently. Use contact.
| What | How often | Why |
|---|---|---|
| Live prices | Daily, automated | Amazon's terms expect prices no older than about 24 hours |
| Roundup picks | Monthly | A pick that has gone out of stock is a broken page |
| Electricity rate | Quarterly | The whole moat is that our number is current |
| Safety content against CPSC recalls | Quarterly | Recalls are the one thing where being stale is dangerous |
| Full content audit | Twice yearly | Specifications change and listings get rewritten |
A note on the name of this page. Inbound links and typed URLs often use the phrase how we test, and that URL redirects here. The redirect exists because it is the industry-standard phrase, not because this page describes testing. It does not, and it never will while the answer is zero.
Sources
- US Energy Information Administration, Electric Power Monthly, Table 5.3 (June 2026): 18.34 cents per kWh, average US residential
- US Consumer Product Safety Commission - recalls database
- UL Solutions - product safety standards and certification marks
Every specification on this page traces to a published manufacturer listing or a government data source. Where a maker publishes no wattage, we say so rather than estimating one.