What this is
,WattDraw computes what household appliances cost to run, using published US electricity prices and a model that accounts for the fact that most appliances spend most of their running time switched off. It exists because the standard published answer to “what does this cost to run” is arrived at by multiplying the number on the label by the hours, and that answer is wrong by a factor of three for the appliances that matter most.
,How it is funded
,Advertising, and affiliate commission on a small number of pages that recommend measurement equipment. Both are disclosed. Neither has any influence on a figure published here, for the straightforward reason that every figure is computed from a public dataset by code you can read the method for — there is no editorial slider to lean on.
,,What it will not do
,- Publish a product review of something nobody has measured. The buying guides here describe what a class of device can and cannot tell you, and where the cheap option is the right one — they do not pretend to hands-on testing that has not happened.
- Invent precision. Where a figure is a typical range rather than a measurement, it says so.
- Change a number because a commercial partner would prefer a different one.
Editorial standards
,Five rules, and they are the reason the site looks the way it does.
,- Every published figure is computed, not typed. Prose that quotes a number reads it from the same data file the calculators use, at build time. A guide cannot cite a price the rate table contradicts, because there is only one source for both.
- Where a figure is a range, it is described as a range. The appliance figures are typical values for typical units under stated assumptions, and every page states its assumption in plain language rather than presenting an estimate as a measurement.
- External checks beat internal consistency. Seven appliances are pinned to regulated published ranges — FTC EnergyGuide, ENERGY STAR, EIA survey means — and the build fails if a computed total leaves its range. A model that only agrees with itself is not evidence.
- Nothing is reviewed that has not been measured. There are no product reviews on this site, because no hardware has been tested here. The buying guides describe what a class of device measures and where the cheap option is the right one.
- The advice that earns least stays on the page. The measurement guide tells you to check your utility’s free interval data before buying anything; the second-refrigerator advice is to retire the appliance rather than replace it.
Where the expertise sits
,This is a data site rather than a personal blog, and its claim to be believed rests on the same thing a dataset’s does: you can check it. The formula is published, the calibration table shows every external check with its computed value beside the published range, and every source file is stored with a checksum so a figure traces to the exact bytes it came from.
,That is deliberate. An assertion of expertise you cannot verify is worth less than an arithmetic you can re-run — and the whole reason this site exists is that a great deal of confidently-written energy advice turns out, when you do re-run it, to be wrong by a factor of three.
,Why this subject
,Because the standard published answer is wrong, and wrong in a way that is easy to demonstrate. Search for what any appliance costs to run and you will find figures produced by multiplying the number on the plate by the hours it is switched on. For a refrigerator that produces about 1,314 kWh a year. The regulated EnergyGuide range for a full-size refrigerator is roughly 350 to 500. The published figure is not slightly high; it is outside the range any refrigerator sold could produce.
,That error survives because it is cheap to make and expensive to correct. Fixing it means knowing that thermostatic appliances cycle, and by how much — which is work, and which no one doing volume content has an incentive to do. It is a good reason for a site to exist.
,How it is built
,A static site generated from two public datasets and one appliance model. There is no database, no content management system and no framework — pages are rendered once at build time and served as files, which is why they load quickly and why every figure on them can be regenerated from source by one command.
,The model that computes every figure is a single file, and the copy that runs in your browser when you move a slider is byte-identical to the one that rendered the page. That is checked on every build. A second implementation in the browser would be free to drift from the printed numbers, and the drift would be invisible — the page would still show a plausible answer.
,How to check any of this yourself
,The strongest thing this site can offer is not a claim to authority but a claim to be checkable, so it is worth saying exactly how. Three levels, each cheaper than the one below it:
,- Check a figure against its published range. The calibration table puts each computed total next to the regulated figure it has to fall inside. If one falls outside, the build is meant to fail; if it did not, that is a bug worth reporting.
- Check the arithmetic. Every appliance page shows its own calculation step by step, with the actual numbers. Multiplying four numbers is not a specialist skill, and if the result differs from the headline the page is wrong.
- Check the source data. Sources names each dataset, its publication date, and a SHA-256 of the file the figures were derived from. Download the same file from the EIA, compare the hash, and you know you are looking at the same bytes.
That third step is the one most sites cannot offer, and it is the reason the raw files are stored here rather than fetched live. A figure traceable to a fixed artefact can be audited years later; a figure fetched from an API at an unrecorded moment cannot.
,Corrections
,Corrections are welcome and are handled in public. A fix goes into the data file the site is generated from, so it propagates to every page that quotes the number rather than being patched on the one page someone noticed. If a figure here is wrong, the fastest way to show it is wrong is to name the source you think is better.
,What “typical” means here
,Every appliance figure on this site describes a typical unit under a stated assumption, and it is worth being precise about what that does and does not claim. It is not the average of all units in service — that average includes twenty-year-old equipment and would describe a machine nobody is buying. It is a mainstream current model in ordinary condition, doing an ordinary amount of work.
,The consequence is that if your appliance is old, tired, or in a hot garage, the figure here is a floor rather than an estimate. That asymmetry is deliberate: it is the direction that leads someone to investigate rather than to relax, and the pages that matter say so explicitly.
,Scope, and what is deliberately missing
,This site covers household electricity in the United States, and nothing else. That is a narrow scope on purpose. Both underlying datasets are US-only, so extending to another country would mean either finding equivalent official data for it or quietly applying American figures to a place they do not describe — and the second is exactly the failure mode this site exists to correct.
,Four things it does not attempt, and will not:
,- Gas, oil and propane. They appear only where a comparison is unavoidable, such as heating. The model is an electrical one.
- Solar, batteries and payback modelling. These need site-specific irradiance and a tariff structure to mean anything, and a national average would be worse than no answer.
- Product reviews. No hardware has been measured here, so none is reviewed here.
- Anything that requires knowing who you are. There is no account system and the calculators run entirely in your browser, which is a design decision rather than a missing feature.
Contact
,Corrections, questions and data suggestions: contact@wattdraw.com. See also how corrections are handled and how the site is funded.