Freya LangfordChief Data Officer · Essays & perspective

From the study / A considered perspective

Hardship and customer data6 min read

Household data should first help the household

Interval data can show an energy retailer that a household is struggling months before a missed bill. Retailers should use that signal only to offer help, build the separation into their systems, and say publicly what the separation costs.

[1]Read the complete essay ↗

On the reading table

  1. [2]

    Forecast the roof, not the region

    Rooftop solar forecasting · 2026-04-21

  2. [3]

    An emissions figure should be reconciled like revenue

    Climate reporting data · 2025-12-03

Customer accounts served, June 2026
1.62 million
1.17 million electricity and 450,000 gas, across four states and the ACT
Disconnections for non-payment, 2025–26
2,900
Down from 6,400 in 2021–22 after early support began
Midday forecast error, residential load
4.1%
Down from 11.4 per cent before the rebuild from customers' meters

A smart meter shows a retailer that a household is struggling long before a bill goes unpaid, and the first use of that knowledge must be to help.

I am the chief data officer of Wirrandel Energy Limited, a fictional ASX-listed retailer and generator based in Melbourne. We supply electricity and gas to 1.62 million customer accounts across Victoria, New South Wales, the ACT, South-East Queensland and South Australia, and we earned A$5.9 billion of revenue in the year to June 2026. Behind the retail book sit two gas peaking stations, a wind farm, a grid battery and a virtual power plant made of about 41,000 batteries in our customers' garages and laundries.

My work begins with the meter. Almost every Victorian home we supply has had a smart meter for more than a decade, and the rollout elsewhere is accelerating. Those meters record what a household draws and exports at intervals of thirty minutes or less. In a retailer's hands that record can price a hedge, settle a battery's contribution to the grid, count the emissions from the gas a family burns, or reveal that a pensioner has stopped heating her house in July. I lead 260 people who build and govern the systems that turn those readings into decisions, and I spend a good part of my week deciding which decisions they should never be used for.

I came to this job by an indirect route: weather statistics, demand forecasting at a Queensland generator, churn analytics at a telecommunications company and five years building collections models at a regional bank. The bank taught me how efficiently data can be used to chase people. At Wirrandel I have tried to build the opposite habit, beginning with a rule that our early-warning signals of payment difficulty go only to the colleagues whose job is to offer help.

Help offered before the first missed bill

Since 2022 our early-support team has contacted 61,000 households flagged by meter and billing signals, and 23,000 accepted a payment plan, a cheaper offer, a missing concession or a counselling referral. Disconnections for non-payment fell from 6,400 in 2021–22 to 2,900 in 2025–26.

A midday forecast built from our own roofs

Rebuilding the residential forecast from 330,000 solar customers' meter exports cut midday forecast error from 11.4 to 4.1 per cent. Our trading team estimates hedging losses about A$19 million a year lower than under the regional vendor model it replaced.

Battery responses we can evidence

We withdrew 6,800 batteries of one make from frequency-control markets for five months, forgoing A$2.7 million, until their records matched independent measurements. Every dispatch event across the fleet is now reconciled automatically, with a match rate of at least 99.2 per cent.

An emissions ledger that reconciles to billing

Our first mandatory climate statement, for the year to June 2026, draws on an emissions ledger reconciled monthly to billed volumes within 0.2 per cent. Building it exposed a 3.8 per cent overstatement in an earlier voluntary figure, which we corrected publicly.

Current focus

I am reviewing every model we run against privacy reforms passed in 2024, which will require organisations to say in their privacy policies how automated decisions that significantly affect people are made. I am also building, with our traders, a five-minute forecast of what our virtual power plant can deliver on summer's hottest evenings.

Conversations I welcome

I am not seeking another executive role. I would consider one non-executive directorship at a household-facing organisation outside energy retailing, such as a water utility, a private health insurer or a customer-owned bank, joining its audit and risk committee. With my board's consent, I could commit about twenty-four days a year. I also accept a few invitations to speak with regulators, financial counsellors and data practitioners.

Beyond the page

The record behind the perspective.

In the boardroom

  • Quorrell Street Financial Counselling LtdNon-executive director · 2022–present
  • Wirrandel Battery Network Pty LtdDirector of the wholly owned virtual power plant subsidiary, in an executive capacity · 2023–present

Formation

  • Bachelor of Science (Statistics and Atmospheric Science)Merrindale University · 1998
  • Master of Applied Statistics, part-timeHalloway Institute of Technology · 2005
Freya brings the committee the test results nobody asked for. Each year she shows us whether the wall between her hardship model and collections actually held, including the two occasions when it was tested and nearly failed. I have sat on boards where data was presented as reassurance. Hers arrives as evidence.
Nerida EllwoodFictional independent non-executive director and chair of the Audit and Risk Committee, Wirrandel Energy Limited
The full record ↗

Contact

I am not seeking another executive role. I would consider one non-executive directorship at a household-facing organisation outside energy retailing, such as a water utility, a private health insurer or a customer-owned bank, joining its audit and risk committee. With my board's consent, I could commit about twenty-four days a year. I also accept a few invitations to speak with regulators, financial counsellors and data practitioners.

  • Board appointments
  • Speaking invitations