Sample invoices
Five fictitious but deliberately realistic documents, day/night meter registers, periods that straddle months, kg instead of tonnes, small print and VAT clutter. Download one, then drag it onto the matching category card on your club dashboard and watch the extraction run.
All suppliers, account numbers and figures are invented. Under each sample: what the engine should pull from it, and why the document is harder than it looks.
Annual electricity statement
Voltara Energy Ltd (fictitious)
A full-year statement with separate day and night meter registers, one row per month, plus a charges breakdown and supplier small print.
- 12 monthly records for 2025, one per month
- Day + night kWh summed into a single monthly figure (e.g. Jan = 934 + 412 = 1,346 kWh)
- March, July and December flagged as estimated (marked E on the bill)
- Standing charges, PSO levy and VAT ignored: they're money, not energy
Why it's tricky: The engine has to know the two registers are the same electricity, not two sources, and that the charges table at the bottom repeats the same kWh and must not be double-counted.
Winter gas bill spanning three months
Hearthstone Gas (fictitious)
A single bill covering 12 Nov 2025 to 09 Jan 2026, 59 days across three calendar months and a year boundary. Metered in m³, billed in kWh.
- The billed 4,342 kWh (gross calorific value, what the SEAI factor expects)
- Pro-rated by days: ~19/59 to November, ~31/59 to December, ~9/59 to January
- All three rows flagged as estimated, because pro-rating is an approximation
- The m³-to-kWh conversion line recognised and not double-counted
Why it's tricky: Billing periods that straddle months (and here, years) are the single most common messy case in utility data. This tests the pro-rating rules directly.
Quarterly waste invoice with a lift log
BinWise Environmental (fictitious)
A Q1 invoice listing eleven individual bin lifts across three waste streams: general (landfill), mixed recycling and organic, with weights in kg per lift.
- Lifts grouped by calendar month and by stream (three subtypes in one document)
- Weights converted from kg to tonnes (the unit waste factors use)
- e.g. January landfill = 128 + 142 = 270 kg → 0.27 tonnes
- Bin rental, levies and VAT ignored
Why it's tricky: One document produces records under three different emission factors. The stream names on the bill ('MSW residual', 'MGB organic') don't match our category names, so the engine maps them across.
Two-month water bill
Regional water utility (fictitious)
A non-domestic water bill for March–April with meter readings, a volume in m³ (and litres), and separate supply + wastewater charges.
- 182 m³ split evenly-ish across March and April by days (31/61 and 30/61)
- Both rows flagged estimated (pro-rated)
- Supply and wastewater charge lines recognised as the same water, not added twice
Why it's tricky: The meter readings contain a typo-style thousands format ('2,4471') exactly like real bills sometimes do. The volume line is the reliable figure.
Coach hire invoice for away fixtures
Shamrock Coach Hire (fictitious)
Five away trips across March and April with round-trip distances and passenger counts per trip.
- Passenger-kilometres per trip (distance × passengers), grouped by month
- e.g. 08 Mar: 148 km × 41 passengers = 6,068 passenger-km
- Hire charges and driver subsistence ignored, cost is not distance
Why it's tricky: Coach emissions are measured per passenger-km, so the engine must multiply two columns rather than lift a single number. An invoice showing only a price would be (correctly) rejected as missing distance data.