The problem it solves
A comparison is only as good as the figures underneath it. If a rate is misread by a few cents, every saving on the page is wrong — and wrong in a way that looks completely convincing, because the arithmetic downstream is flawless. Confident and wrong is the worst outcome this product can produce.
So each bill is checked against itself before anything is shown to you.
How it works
- The rates, usage, supply charge, controlled load and solar export are read off the bill.
- Those figures are used to rebuild what the bill should have totalled for its own period.
- That rebuilt figure is compared to the total the bill actually states.
- If they agree closely, the reading is verified against the document itself and the comparison proceeds on trusted figures.
- If they do not, that is surfaced rather than hidden, and the reading is attempted again.
Why the check is allowed to fail loudly
It would be easy to make this check always pass — adjust a rate until the arithmetic closes. That is precisely what it must never do. A bill reconstructed from figures that were never printed on it reproduces perfectly and is completely fabricated, and the check would then certify the fabrication.
When a reading cannot be verified, the second attempt is told what did not add up but is never told the size of the gap or the total to hit, and is explicitly instructed not to adjust any figure to make it balance. A reading that does not add up is more useful than one that adds up because a number was changed.
The second reading is only used if it scores better than the first on an independent quality check. A retry can improve the result or be discarded — it cannot make it worse.
What is checked beyond the total
- Rate plausibility
- A usage rate of 0.32c or 320c per kWh is a decimal-point slip, not a tariff. Implausible figures are dropped rather than used.
- Bucket consistency
- Usage recorded against one tariff structure with rates recorded against a different one cannot be priced, and if left alone would silently charge your whole consumption at the peak rate on every time-of-use plan.
- Controlled load completeness
- Consumption without a rate, or a rate without consumption, means a charge is missing from the reconstruction.
- Solar consistency
- A feed-in credit larger than every charge on a bill that still says money is owed cannot be right, whatever the individual figures look like.
- Network identification
- A bill with no distribution network returns no plans at all, so the network is derived from the postcode where that is unambiguous — and left for you to confirm where it is not.
Why percentages can mislead on solar bills
If you have solar, your bill is a large credit netted against large charges, and the total left over can be small. A modest error in the gross charges then shows up as an enormous percentage error on that small net figure — a real bill reconstructed 40% out on a $12 total, which was an error of $4.88.
So reading quality is tracked in dollars as well as percentages, and solar-netted bills are measured separately. Judging them on percentage alone would send you chasing the wrong problem.
Frequently asked questions
What happens if my bill cannot be verified?
You are told. The comparison still runs, because the bill total itself is usually correct, but the claim that we reproduced your bill is withheld rather than asserted.
Does a failed check mean my saving is wrong?
Not necessarily — the headline saving is measured against the bill total, which is usually read correctly. What is less certain is the per-rate detail.
How accurate is the reading in practice?
It is measured, not assumed: extraction is scored against a set of real bills, and against every bill already stored, so a change can be shown to have improved things rather than argued to have.