From test cards to telemetry terms

Mandate

Establish accountability for machine, product, quality and customer data across engineering, the plants, service and dealers.

Outcomes

• Published grower data terms separating machine-health, agronomic and location records; telemetry renewals rose from 81 to 90 per cent
• Moved a failure model to pre-season stocking, lifting same-day drive-kit availability from 69 to 91 per cent
• Introduced first-season warranty claims per hundred machines as the shared quality measure for Toowoomba and Dalby
• Agreed governance rules for the Dalby row-unit camera check with workforce representatives

Decisions in context

The chief data officer role was created after the fund's directors asked who answered for data that crossed engineering, the plants, service and dealers. My first task was to write down owners, and several colleagues found that list uncomfortable because it gave them accountability they had previously shared with others. The data office now has four teams: data engineering, analytics, master data, and governance and privacy. We work under a charter approved by the audit and risk committee, to which I report twice a year. The grower data terms and the redirected failure model both came from this period, and each involved giving up something a colleague valued. I hold no authority over design, safety or warranty acceptance. My contribution is to make the records reliable and to set the rules under which others may use them.

Reason for the move

Current appointment; seeking a first non-executive role alongside it, with employer consent.

Mandate

Build an analytics capability and the data foundation for machine telemetry.

Outcomes

• Brought data from 2,800 connected machines into a single warehouse by 2020
• Retired 21,500 duplicate or superseded part numbers; wrong-part returns fell from 7.4 to 3.2 per cent
• Closed a branch dashboard programme after usage records showed it was rarely consulted

Decisions in context

I started with four analysts and a request from the managing director to explain why our reports disagreed with each other. Telemetry arrived in 2018, first on planters and then on sprayers, and the data quickly outran our agreement about who could use it. Much of my time went on foundations that rarely appear in presentations: one register of customers and machines, reconciled serial numbers and a warehouse that engineering and service both trusted. The parts catalogue clean-up belongs to this period, as does a decision I still describe to new team members. We had built a dashboard for branch managers, and usage records showed that most opened it fewer than twice a quarter. I closed it and asked the managers what they needed to decide each week. Their answer was three numbers, sent by email on Monday morning.

Reason for the move

Appointed the company's first chief data officer when the board created the role.

Mandate

Make released engineering data accurate enough to support parts ordering and dealer service.

Outcomes

• Linked each released bill of materials to its parts-catalogue page
• Introduced a weekly catalogue change file for dealers' ordering systems
• Reduced catalogue publication time for a new machine from eleven weeks to four

Decisions in context

This role taught me that data quality in manufacturing is decided long before anyone runs a report. My team took each engineering release and turned it into the bills of materials used by the plants and the illustrated catalogue used by dealers. When engineering changed a part without recording what it replaced, the consequence surfaced months later at a dealer's counter. I spent many hours with the engineering director agreeing which changes needed a supersession decision and who would make it. We did not always agree, and engineering held the final authority over design. My job was to show the downstream cost of an incomplete release in terms the design team could weigh against its schedule. I also learnt to respect the technical illustrators, whose drawings dealers trusted more than any written description, and that respect for the end user of a record has stayed with me.

Reason for the move

Asked by the managing director to form the company's first analytics team.

Mandate

Turn dealer warranty claims into evidence engineering could use to improve designs.

Outcomes

• Replaced free-text claims with a coded failure-mode form and a photograph requirement
• Cut average claim assessment time from 26 days to nine
• Identified a chaser-bin auger gearbox failure pattern that led to a design change in 2010

Decisions in context

I joined Callowdene to make warranty claims useful. Dealers wrote claims in their own words, and a failed part might arrive with no tag, a superseded number or a description that said only that it broke. My six assessors judged each claim against warranty policy, and I took the patterns to engineering. Dealers resisted the coded claim form at first, seeing it as more paperwork for the same payment. Paying coded claims faster changed that view within two seasons. The most important finding of the period was a gearbox failure on our chaser-bin augers, which the claims showed clustered in machines handling heavier harvest loads. Engineering redesigned the gearbox mounting in 2010. I did not decide the redesign; I supplied evidence that made the case for it difficult to dismiss, and that became my working definition of a data role.

Reason for the move

Moved to lead engineering data and parts information as the product range expanded.

Mandate

Reduce unplanned crushing stoppages through better use of maintenance and condition data.

Outcomes

• Linked vibration and oil-sample results to the slack-season work list
• Helped reduce unplanned crushing stoppages from mill-drive gearboxes by about a third over three seasons
• Earned a master's in applied statistics through evening study alongside the role

Decisions in context

Sugar milling runs on a fixed calendar. During crushing, every stoppage means cane deteriorating in bins and growers' deliveries backed up at the siding. In the slack season the plant is opened up and rebuilt. My early predictions arrived at the wrong point in that cycle and changed nothing. I learnt to present condition evidence while the shutdown scope was still being drafted, ranked by the cost of an in-season failure rather than by statistical confidence. The maintenance superintendent made the final call on the work list; my role was to make the trade-offs visible to him. Over three seasons, unplanned stoppages from mill-drive gearboxes fell by about a third, although better maintenance planning deserves a share of the credit. The co-operative's grower-members also taught me to ask whom a shared asset serves, a question that returned years later with telemetry.

Reason for the move

Joined an agricultural machinery manufacturer to work on product reliability rather than plant reliability.

Mandate

Run production performance tests and maintain reliable test records.

Outcomes

• Added test conditions to every performance record
• Traced an output shortfall on one pump model to a single impeller casting batch
• Wrote the test procedure for the company's larger centrifugal models

Decisions in context

My first employer made irrigation and stock-water pumps, and my first responsibility was its test bench. Every production pump was run and its flow and pressure recorded on a card. Within a few months I noticed that results for one model varied more than its design could explain. Recording water temperature, gauge calibration date and impeller casting batch beside each result showed that one batch of castings produced pumps about six per cent below rated flow. The engineering manager took the matter up with the foundry; I supplied the evidence. The work was modest, but it established the habit of capturing the conditions of a measurement along with the measurement. I also learnt to talk to the people who used the result: sales staff quoting performance to growers, and service technicians who heard when a pump fell short.

Reason for the move

Moved into plant reliability to work with larger equipment and condition data.