Thirty years between the ledger and its users

Mandate

Modernise the ledger, channels and controls while reducing the manual work that systems impose on operations, and report technology and cyber risk to the board.

Outcomes

• Opened a workaround register of 212 entries and reduced it to 104 by June 2026
• Deferred and narrowed the core replacement to 46 interfaces before signing
• Introduced source verification of income on every broker mortgage file
• Moved payment exception handling from an overnight team to automated intraday posting

Decisions in context

Returning to a company whose ledger I once programmed gave me an unusual advantage and a risk. I knew why parts of the system behaved as they did, but I also had to resist defending choices I had helped make in 2003. I began with the register rather than a strategy document, because the people closest to the work could tell me more than any architecture review. Five directors lead core platforms, digital channels and integration, infrastructure and cloud, information security and the delivery office. I own the technology budget and the vendor relationships, while operations, credit and the estate business own the processes my team changes. The three cases on this site each involved a colleague with a strong opposing argument. I present to the risk committee and to the board for major approvals; I am not a director.

Reason for the move

Current appointment; seeking a first compatible corporate board seat with employer consent.

Mandate

Run and improve the shared core banking platform for member credit unions serving about 620,000 members.

Outcomes

• Published each member credit union's manual reconciliation count beside its feature requests
• Consolidated three card-processing arrangements into one contract
• Moved member-facing online banking off the core's batch statement cycle

Decisions in context

Tellmarsh taught me to serve owners who disagreed. Some member credit unions wanted mobile features their competitors already offered; smaller ones wanted the platform to stop creating reconciliations their two-person back offices had to complete by hand. Both requests were reasonable and the budget could not satisfy both in the same year. I began reporting each member's manual reconciliation count beside its feature requests and took the trade-offs to the board technology and operations committee instead of settling them inside my team. The committee did not always choose as I would have, which was the right outcome for a cooperative. I also renegotiated the card-processing arrangements and moved online banking away from the batch statement cycle. Watching that committee work gave me my first sustained view of directors deciding among legitimate interests.

Reason for the move

Returned to Pembrevale Trust Company as chief technology officer.

Mandate

Connect client institutions' batch ledgers to payment networks that expected answers in seconds.

Outcomes

• Standardised the integration kit used by new client institutions
• Counted next-morning repairs at client sites as part of each interface's cost
• Reduced unmatched electronic funds transfer items for the largest clients

Decisions in context

At Quintrel I worked between two clocks. Payment networks wanted authorisations and confirmations within seconds; most of our clients' ledgers settled their books overnight. My team built and maintained the interfaces that held those timetables together. The technical work was demanding, but the more revealing evidence came from client operations analysts who repaired failed items each morning. I started recording those repairs by client and interface, and our product reviews began treating them as a cost Quintrel was passing to its customers. That changed which integration improvements we funded first. I also learnt how contract language allocated responsibility for an item that failed between two systems, and how often neither party had planned for the person who would actually resolve it. It was the first time I counted manual work deliberately rather than noticing it by accident.

Reason for the move

Moved to Tellmarsh to run a core platform rather than connect to other institutions' platforms.

Mandate

Deliver releases of the mortgage-servicing product that clients could adopt without local modifications.

Outcomes

• Introduced back-office visits to clients as part of release planning
• Added broker-data validation that removed several client-specific adjustments
• Reduced the number of client customisations carried into each release

Decisions in context

Selling one product to many lenders showed me how differently each used it. Our clients processed renewals, payment changes and prepayments with a mixture of our software and their own adjustments, and those adjustments rarely reached our release plan. With our product managers, I began visiting client back offices to watch the work happen. Several popular feature requests were really workarounds for incomplete broker data, which meant the cleanest fix was validation at intake rather than another screen. My sixteen developers and testers worked to a fixed release calendar, and I learnt to decline custom work that would have helped one client while making every future upgrade harder for all of them. The habit of watching a process before designing for it began here, as did my understanding of mortgage servicing from the lender's side.

Reason for the move

Joined a payment processor to work on real-time interfaces between institutions.

Mandate

Maintain and change the in-house COBOL deposit ledger that calculated GIC interest, maturities and renewals.

Outcomes

• Rewrote the GIC renewal notice process that clerks had corrected by hand
• Documented interest calculation rules for audit and estate staff
• Led three programmers on maturity and renewal changes

Decisions in context

My first period at Pembrevale gave me the ledger from the inside. I maintained COBOL programs that calculated GIC interest, maturities and renewal instructions, and I took change requests from deposit clerks, estate officers and accountants. Many requests that looked small were really attempts to stop a manual step someone performed daily. The renewal notice rewrite was one: clerks had been correcting printed notices by hand for years. From 2004 I led three programmers, which taught me to explain priorities to colleagues outside technology. I had no budget authority and did not choose what the department funded. I left after five and a half years because I wanted to understand how lending systems worked across many institutions rather than one, not expecting to return fifteen years later to decide what replaced the programs I had written.

Reason for the move

Moved to a software vendor to see lending systems across many institutions.

Mandate

Maintain overnight batch processing and remediate date handling before 2000.

Outcomes

• Remediated two-digit year handling across eleven credit union ledgers
• Recorded manual corrections that client staff made to bureau reports
• Rebuilt the overnight job schedule documentation

Decisions in context

The bureau ran other organisations' books overnight on a shared mainframe, and I started by monitoring jobs and fixing failed runs before the morning. Year-2000 remediation then gave me three months inside eleven credit union ledgers, reading every date field. At one credit union I found a binder of seventeen corrections staff made by hand each morning because one of our reports had never matched the ledger. Nobody had reported it; they had simply adapted. I recorded those corrections for my supervisor, and several reports were rewritten. I held no authority beyond my assigned jobs, but the episode gave me the question that still shapes my work: what are people doing around this system that its owners cannot see? I joined Pembrevale in 2000 to work closer to the users of a single institution's ledger.

Reason for the move

Joined a trust company to work on a single institution's ledger and its users.