Your utility meters you every hour and must hand you the file. This reads it, shows you your own habits, and then re-bills a year of them under every rate plan you could be on — or tells you exactly why it will not. No account. No upload.
Drop your usage file here
or click to choose · a .csv from PG&E or SDG&E
Your file never leaves your computer. The analysis runs inside this page; there is no server to send it to — exactly what the page does request.
What comes back. Your load shape, then a ranking of every plan you could be on — priced for PG&E households this engine has been proved against, refused by name for the rest. Nothing here guesses.
The taller the bar, the more electricity you use in that hour. The shaded band is 4–9 p.m., the most expensive window on both PG&E's E‑TOU‑C and SDG&E's TOU‑DR1 — the two utilities' standard residential plans. Usage inside the band costs the most, which is why this one chart matters more than the rest.
Seasons change what you use, and a plan judged on winter alone will be judged wrong. Partial months at either end of your file will look artificially low.
Nothing above means anything if the file was misread. This is exactly what was found in it.
Everything above is arithmetic on your own meter readings, which is safe to compute for anybody. Turning it into money means modelling a specific tariff, and the project holds itself to a rule before any dollar figure ships:
That is met for PG&E — eleven consecutive real statements reproduced, worst error 22 cents. It is not met for SDG&E: no real SDG&E bill has ever been tested against the engine, so San Diego files get the analysis above and no price, full stop.
Passing that gate is necessary and not sufficient. The rate specs describe a particular household — one baseline territory, one heat source, one generation supplier, one PCIA vintage. A PG&E customer outside those is refused too, by name, in section 03, because the same arithmetic run against the wrong allowance is not approximately right — it is wrong by real money. The full method is written up here.
This is the difference between a tool that is independent and one that is selling you something: the refusal is the product. A number here would be cheap to produce and impossible for you to check. How the ±$2 rule works.
Annual cost by plan, cheapest first
A ranking with no explanation is a number to take on faith. These are the bill lines that actually separate the two plans, largest first.
Free, about two minutes, and it is your data by law. Utilities call it Green Button.
Nothing is uploaded and nothing is stored. Close the tab and it is gone.
When you use power, how much, and whether it lands in the expensive hours.
Every eligible plan re-billed on your intervals, cheapest first, with the usage shift that would overturn the answer — or a plain statement of why this engine has not earned the right to price your household.
You want the interval usage export — the hour-by-hour readings. The billing or cost summary is a different download and will be turned away here.
pge.com and open Energy Usage Details.sdge.com and open your usage or
Green Button tool.If you have solar, your file records what you sent to the grid separately. This reads that too, and will say so.
California rates change between summer and winter, and so does most households' usage. A comparison built on three months of winter data will mis-rank plans for the summer you never measured.
Both utilities keep 13 months. Export all of it while you can — access usually ends when the account does.
The code running in your browser is the same Python the project's test suite covers, compiled to WebAssembly — not a separate copy rewritten for the web that could quietly drift out of step. The bill engine behind the wider project is held to a published accuracy record against real statements.
Every tariff number in the engine cites the utility tariff sheet it came from; an unknown value raises an error rather than quietly defaulting to a guess. Read the methodology