Six appointments, from Queensland tills to a listed retailer's technology

Mandate

Replace an ageing commerce platform without losing a peak season, and bring cyber security and customer-data handling to the standard now expected of a listed retailer.

Outcomes

• Retired the 2009 commerce application through 31 reversible releases over 34 months for A$71 million; no Black Friday or Boxing Day outage has exceeded four minutes since 2022.
• Captured A$5.8 million of orders in deferred-authorisation mode during a 47-minute gateway failure on Black Friday 2024, with confirmed fraud of A$41,000.
• Deleted 1.7 million stored identity-document images and de-identified 4.6 million dormant customer records, cutting worst-case breach exposure from about 6.1 million people to 1.9 million.
• Launched a customer-service assistant that resolves 31 per cent of chat and messaging contacts and saves about A$3.6 million a year.
• Raised online conversion from 2.1 per cent to 2.6 per cent and cut 95th-percentile page time from 3.8 seconds to 1.4 seconds.

Decisions in context

Halvorne created the chief technology officer role after its website failed for three hours and ten minutes on Boxing Day 2019; until then technology had reported to the chief financial officer. I joined in April 2020, a few weeks into the pandemic, and the first year was spent keeping an old platform alive while online orders doubled and, during Melbourne's stricter lockdown, briefly carried most of the business. The board then approved a tranche-funded modernisation rather than a single cutover, and we retired the old application in August 2023. Since then the work has turned to payments resilience, deletion of customer data the business no longer needs, a deliberately narrow customer-service assistant and the replacement of point-of-sale systems in all 156 stores. I have seven direct reports, including a chief information security officer with a direct line to the audit and risk committee chair.

Reason for the move

Current role. Not seeking an executive move before the store-systems programme finishes in mid-2027; considering one non-executive directorship with the Halvorne board's consent.

Mandate

Stop the marketplace failing during its sale events and build the payment, payout and seller-verification systems a growing marketplace needed.

Outcomes

• Kept the marketplace available for 99.97 per cent of its two largest sale weekends in 2018 and 2019, after failures in two consecutive events before I joined.
• Moved hosting from one region to two, with failover rehearsed every quarter.
• Moved seller payouts from weekly to daily and introduced identity and bank-account verification for all sellers.
• Reduced fraudulent chargebacks from 0.84 per cent to 0.31 per cent of payment value.

Decisions in context

Pellbrook connected about 3,800 independent sellers with shoppers looking for new and refurbished goods, and it had failed during two consecutive sale events before I arrived as head of engineering in August 2016. The cause was less the traffic than the design: every page depended on a single seller-inventory database that locked under load. We split that dependency, introduced a waiting room and moved hosting to two regions with rehearsed failover. When a private equity firm took a majority stake in July 2018, I became general manager of technology and began presenting to the board. That board wanted growth and was entitled to ask what each risk would cost, which taught me to bring load-test results rather than opinions. We also paid sellers daily instead of weekly, verified their identity and bank accounts, and cut fraudulent chargebacks by almost two-thirds.

Reason for the move

Wanted to run all of technology for a retailer with stores, where the website, the stock and the counter have to agree.

Mandate

Make the merchant gateway reliable enough for large retailers and reduce the card data merchants had to hold themselves.

Outcomes

• Raised gateway availability from 99.86 per cent to 99.99 per cent over three years.
• Built a card tokenisation vault adopted by 2,100 merchants, removing stored card numbers from their systems.
• Passed the annual PCI DSS service-provider assessment in each year of my tenure.
• Offered 3-D Secure to online merchants, roughly halving fraud chargebacks among those who adopted it.

Decisions in context

Returning from England, I joined Corrandel in March 2012 to run engineering for its merchant gateway, which then processed card payments for about 6,800 online and in-store merchants. Large retailers were beginning to ask us for evidence of availability rather than assurances, and our record did not yet support the conversation. Over three years we removed single points of failure, built active recovery in Melbourne and raised availability from 99.86 per cent to 99.99 per cent. I became head of payments engineering in 2014, reporting to the founder. The tokenisation vault we built let 2,100 merchants stop storing card numbers, which shrank their compliance burden and their exposure in a breach. The most lasting lesson came from the provider's side of an outage: our status page always lagged our merchants' failing checkouts, and their customers never cared whose fault it was.

Reason for the move

Wanted to work for a merchant and own the customer's outcome, not only the gateway's uptime.

Mandate

Build order-capture features for the online store, then design the migration of catalogue order entry after the failed 2010 cutover.

Outcomes

• Led the recovery workstream that restored stable order capture after eleven days of intermittent failure in November 2010.
• Designed a phased migration of catalogue order entry in nine releases, completed in January 2012 without a peak-season outage.
• Introduced daily reconciliation of every order between the old and new systems during migration.

Decisions in context

I moved to Leeds in October 2008 as a senior developer on Kelderby's online store, at a time when the company was moving customers from paper catalogues and telephone orders to the web. In November 2010 the business replaced its order-taking platform in a single cutover seven weeks before Christmas. Order capture failed on and off for eleven days, and the old platform could not be restored because its data had already been migrated and changed. I led the workstream that stabilised order capture, working nights with the contact centre to re-key orders that had been lost. The IT director then asked me to design the migration of catalogue order entry, and I did it in nine reversible releases with daily reconciliation of every order. That programme finished in January 2012 without an outage, a month before I returned to Australia.

Reason for the move

Returned to Australia when my partner accepted a role in Sydney.

Mandate

Maintain and extend till software across 31 stores, then make it dependable in stores with unreliable network links.

Outcomes

• Built an offline till mode that kept stores trading and sent sales to head office when links returned, deployed to all 31 stores by 2007.
• Replaced a paper lay-by system with a till-based one, roughly halving lay-by disputes.
• Replaced store-by-store software visits with overnight remote updates for every till.

Decisions in context

Varrowby was a family-owned department-store group with 31 stores from the Gold Coast to the far north, and I joined its small development team in February 2004 to maintain the till software. The tills had been designed on the assumption that head office was always reachable. In the 2006 wet season, storms cut the links to our northern stores for days at a time, and staff were reduced to writing sales on paper. I led the rebuild of the tills so that they could keep trading offline, hold their records safely and send them to Brisbane once a link returned. By 2007 every store had it. I also replaced the paper lay-by system, which customers relied on before Christmas, and introduced remote overnight updates so that a software release no longer meant a technician driving to each store.

Reason for the move

Wanted to see how larger retailers ran catalogue and online businesses, and to work overseas for a few years.

Mandate

Build and support interfaces between retailers' stock, point-of-sale and supplier systems.

Outcomes

• Wrote stock-replenishment interfaces for four regional retail chains.
• Built an invoice-matching service that a freight client used for about 9,000 invoices a week.
• Progressed from graduate to developer in 2001 and led the development work on two client projects by 2003.

Decisions in context

Oxmere was a Brisbane integrator of about 160 people, and I joined as a graduate developer in February 2000, two months after finishing my engineering degree. Its work was connecting systems that had never been designed to meet: a regional retailer's stock files and its suppliers' order systems, a freight company's invoices and its customers' accounts. I wrote stock-replenishment interfaces for four regional retail chains and an invoice-matching service that a freight client ran on about 9,000 invoices a week. The work taught me that most failures happen at the joins, where each side assumes the other has checked the data. By 2003 I was leading the development work on client projects and mentoring two graduates. I left because I wanted to stay with a system after it went live and live with the consequences of my own design.

Reason for the move

Wanted to work inside a retailer and own a system after it went live rather than hand it over.