From a mine's monitoring centre to the data office

Mandate

Run one accountable data function for trading, retail customers and charging, and meet personal-information and data-security obligations across all three.

Outcomes

• Installed gateways at 44 large customer sites, cutting weighted day-ahead load error from 8.7 to 4.6 per cent
• Declined a session-level data sale and reduced identifiable charging records held from 260 million to 31 million
• Linked charger alarms to work orders, leading to faults confirmed at 39 connectors before the 2025 holiday peak
• Began writing data-use terms into retail contracts at renewal

Decisions in context

As chief data officer I answer for every dataset the company holds, from fifteen-minute customer readings to the phone numbers in our charging app. Each dataset has an owner in the business; my office sets the rules for its purpose, quality and retention, and supplies the analytics that trading, customer service and field operations depend on. I present quarterly to our board and its risk committee on data security, personal information and the cost of forecast error. The appointment has asked me to say no more often than I expected: to a data sale, to analysis requests without a stated purpose, and to my own preference for trading work over safety records. All three cases here come from this period. I hold no directorship, and I would treat a first board appointment as a role to learn rather than an extension of this one.

Reason for the move

Current appointment; considering one compatible independent board role.

Mandate

Combine the data of the retail business and an acquired charging operator without merging their governance.

Outcomes

• Moved retail and charging data onto one platform over thirty months
• Created a separately governed store for drivers' personal information
• Named an owner for each of 140 core datasets

Decisions in context

The charging acquisition doubled the systems I was responsible for and changed the nature of the records. Retail customers are companies with contracts; drivers are individuals using an app, and their records carry personal information. I led a 22-person team that moved both businesses onto one platform while keeping drivers' records in a separately governed store with its own access rules. That separation slowed some analyses and frustrated colleagues who wanted to combine charging behaviour with commercial customer data. I held to it because the purposes differed. We named an owner for each of 140 core datasets and set retention periods for most of them. Charging sessions stayed marked as undecided, and I let that stand too long. The work taught me that a platform programme is mostly a series of decisions about who may use what, and only partly a choice of technology.

Reason for the move

Appointed chief data officer when the board created the role to answer for data security and personal information across the company.

Mandate

Build the company's first customer load forecasts and prepare analytics for the provincial spot-market trial runs.

Outcomes

• Built interval load forecasts for 420 customers from delayed metering data
• Prepared settlement analytics for the spot-market trial runs
• Introduced forecast ranges and a written record of trader overrides

Decisions in context

I joined a retailer with a registration, a small trading desk and customer data held in spreadsheets. Consumption arrived through the grid company's metering service a day or more late, and some customers sent production plans irregularly. With a team of six I built forecasts for annual and monthly contracts, then prepared settlement analytics for the province's spot-market trial runs. The habit that mattered most came from the coal plant: show how far a forecast can be trusted before anyone acts on it. We published ranges rather than single lines and asked traders to write down when they overrode us. Some traders found that tiresome. Over time the override record became our most useful learning tool, because it showed when their judgement beat the model and when it did not. I reported to the head of trading and had no authority over positions.

Reason for the move

Given responsibility for data across the company after the 2019 charging acquisition.

Mandate

Make fuel, continuous emissions monitoring and early carbon-reporting data consistent and traceable.

Outcomes

• Wrote a chain-of-custody procedure from coal sample to laboratory result
• Reconciled continuous-monitoring records with unit operating logs every month
• Prepared the plant's first carbon-emissions reports

Decisions in context

Supervising fuel and emissions data meant answering to two audiences with different questions. Plant management wanted the fuel account to close; the environmental team needed monitoring records that matched what the units had actually done. With nine colleagues I reconciled continuous-monitoring data against unit operating logs each month and investigated gaps before they were filled. The carbon reports were new work. Their accuracy depended on coal sampling, laboratory analysis and fuel consumption figures that had previously served only cost control. I wrote a chain-of-custody procedure so that every laboratory result could be traced to a sample, a date and the person who took it. I reported findings to the plant's safety and environmental committee, which was my first regular experience of presenting uncomfortable numbers to a formal body and answering for them.

Reason for the move

Joined electricity retail after the 2015 reforms opened it to new companies.

Mandate

Reconcile coal receipts, stock and consumption, and report unit efficiency each month.

Outcomes

• Explained recurring gaps between weighbridge, stockpile-survey and burned-coal figures
• Built monthly heat-rate reports for four units
• Documented how moisture and sampling practice affected the fuel account

Decisions in context

My work was to make the fuel account explain itself. Coal arrived by rail and road, was weighed, sampled, stockpiled and burned, and at each step the quantity was measured differently. A monthly gap between purchased and burned coal could reflect moisture loss, a stockpile survey, sampling practice or a genuine loss, and each called for a different response. I learnt to present those sources separately rather than offer one unexplained difference. I also built monthly heat-rate reports for the four units, which the performance engineers used to plan maintenance. I had no direct reports and no authority over operations. The role taught me patience with physical measurement and a lasting suspicion of any total that hides how its parts were obtained, a habit that later shaped how I treated both customer meter data and charger telemetry.

Reason for the move

Promoted to supervise fuel and emissions data.

Mandate

Summarise gas-monitoring readings and alarms with the calibration and outage context the safety engineers needed.

Outcomes

• Combined alarm logs with calibration records in one weekly report
• Flagged sensors overdue for calibration before their readings were relied on
• Recorded monitoring-line outages so that gaps were never read as safe readings

Decisions in context

The monitoring centre watched methane, carbon monoxide and airflow sensors across two underground mines, and every alarm was printed and filed. My task was to turn those logs into something the chief safety engineer could act on. The first lesson came quickly: a reading meant little without knowing when its sensor was last calibrated and whether its line had been working. I combined the alarm log with calibration sheets and outage records in a weekly report, and flagged sensors whose readings should not be relied on until checked. I had no authority over ventilation, production or evacuation; those decisions belonged to the engineers. What I gained was a lasting respect for the people who act on data under pressure, and an understanding that a missing reading is not the same as a safe one.

Reason for the move

Moved to power generation to work on fuel and plant performance data.