What an energy retailer's data function is actually for

Interval data is the most revealing record a retailer holds. It shows when a household wakes, whether anyone is home by day, when a shift worker sleeps and when a heater goes off to save money. I run Wirrandel's data governance on the principle that every use must serve the customer or meet a clear legal obligation, and that a new use is a new decision rather than an extension of an old one. The Consumer Data Right now lets a customer direct us to share billing and usage data with an accredited recipient, and we answer those requests from the same governed source that produces the bill, so a comparison service sees exactly what the customer sees. We do not sell meter data. Proposals to share it with partners go to a data ethics panel that I chair, with our privacy officer, a customer advocate and an external financial counsellor whom we pay for her time.

Retailers must assist customers in payment difficulty, through hardship policies approved by the AER in most of the states we serve and through Victoria's payment difficulty framework. Those obligations describe what we owe a customer once difficulty is known; they say less about noticing it. Our early-support model combines meter, billing, concession and payment data to flag households whose pattern has changed: winter use falling in a home with a concession card, instalments shrinking or slipping, bills rising faster than usage explains. Its output goes to a 38-person early-support team and nowhere else. Collections and credit systems cannot read it, and it is excluded from marketing. Customers on the life-support register are handled through a separate protected process that the model never alters. I judge the model by help accepted, disconnections avoided and hardship debts cleared, never by money recovered.

In South-East Queensland and South Australia more than four in ten of our households have rooftop solar, and on a mild spring day our residential load at one o'clock can fall below its level at three in the morning. A retailer hedges expected load in advance, so a forecast that underestimates midday solar buys protection against demand that never arrives. My forecasting team builds the residential forecast from each solar customer's meter exports, inferring every system's effective size and orientation and combining those estimates with irradiance and cloud forecasts by postcode. We report error at midday, at the evening peak and on the ten hottest days of the year, because a monthly average hides the days that cost money. An outside reviewer engaged by our chief risk officer validates the model each year, and the trading desk sees every change before it goes live.

Our virtual power plant enrols about 41,000 home batteries from three manufacturers, with a combined discharge capability of roughly 200 megawatts. We dispatch them into evening price spikes and offer part of their capacity into frequency-control markets, which pay for fast responses to disturbances on the grid. Each use creates a data obligation: evidence for the market operator that the response happened, accurate credits for the customers whose batteries did the work, and a true position for our risk team. I insist that our records rest on measurements we can verify, not only on manufacturers' cloud dashboards. Every event is reconciled against independent measurements, and a battery whose response cannot be evidenced is withdrawn from the market until it can be. Each customer's chosen backup reserve is protected in the dispatch rules themselves: no instruction can draw a battery below it.

Australia's mandatory climate-related financial disclosures took effect for the largest companies for financial years beginning in 2025, so Wirrandel's first statement covers the year to June 2026. It forms part of the annual report, the directors are responsible for it, and our auditor's assurance over it widens in stages. For a gentailer the largest emissions are not our own stations but scope 3: the gas our customers burn and the generation behind the electricity we buy and resell. I led the decision to build the emissions ledger on our enterprise data platform, calculated account by account from billed volumes, with the same reconciliation, change control and sign-off as revenue. The ledger reconciles to billing within 0.2 per cent each month. Facility emissions from our peaking stations still come from the measured data we have long reported under Australia's National Greenhouse and Energy Reporting scheme.