What the insurance file taught me about technology leadership

I grew into technology leadership by watching experienced insurance colleagues explain files that a system had simplified too far. After completing my information systems degree at Lysmere College in May 1990, I joined Vellorin Assurance in September as a systems analyst. I worked in a ten-person team and had no direct reports. My assignments concerned policy administration and records passed into claims. I learned to ask which version of a document someone had read before discussing whether the program had processed it correctly.

The insurer was a privately owned regional business with 460 employees and USD210 million annual direct written premium when I joined. I was responsible for analysis and testing, not for choosing policy terms or settling claims. One early project involved sixteen document types whose labels looked interchangeable to our development team. Claims colleagues showed us that the timing and purpose of each document mattered. I began keeping examples of situations in which a technically valid record could still mislead its reader. That habit became more useful than any particular software language.

In July 1998 I moved to Caldermere Mutual to manage policy platforms. The mutual had 700 employees and USD420 million annual direct written premium. I led twenty-six people through four direct reports, with a USD6 million application budget. Independent agents needed dependable transactions, while service colleagues needed to understand what had happened when a member called. I learned that a release was not complete because the installation had finished. People had to be able to explain the change, locate the relevant evidence and deal with an awkward exception without inventing a private workaround.

My next appointment, director of claims technology in January 2006, put those principles closer to a difficult human moment. I led fifty-four colleagues through five direct reports and managed USD14 million of technology expenditure. The chief claims officer owned claim decisions; I owned the delivery of the supporting systems. We sometimes disagreed over how much information a screen should require. A mandatory field could improve consistency in normal work and obstruct help after a severe loss. I learned to distinguish an incomplete file from a file that could not reasonably be completed yet.

In July 2012 I became vice president for enterprise delivery, responsible for eighty-eight employees and a USD29 million annual programme portfolio. Underwriting, finance, claims and agent service all had credible demands. The hard conversations concerned sequence and organisational capacity, not whether technology mattered. I began asking sponsors to name the work that would disappear, the work that would be added and the person authorised to make that exchange. Several attractive business cases became less attractive when training, checking and assisted service were described honestly. That was useful progress even when it postponed an investment.

I became CIO in October 2018. Caldermere now serves eight Midwestern states, with 2,400 employees and USD1.82 billion direct written premium in 2025. My remit covers architecture, applications, technology operations, delivery and the stewardship of information services. I lead 186 employees through seven direct reports; sixty contracted specialists are counted separately. The chief executive is my reporting line. I present to the board risk committee, but a presentation is not a director appointment and does not make me the owner of every risk discussed in that room.

The recent claims-drafting pilot is a useful example of my approach. Early testing against closed files found that a fluent letter could cite an outdated endorsement or a reference from the wrong jurisdiction. I supported a narrower tool that prepared material for an authorised colleague to review. This gave up a larger automation claim and retained a checking cost. In the later pilot, eighty-three of nine hundred letters were rewritten before release. That finding did not establish a failure rate for every possible claim; it showed why the review step still mattered in the work we had examined.

I also care about colleagues and members who do not fit the main journey. When fifteen hundred of seventy-six hundred invited members used telephone or post for renewal, we kept an assisted route and measured its cost. Calling those members digitally unsuccessful would have hidden a service choice we still had to make. The technology team could simplify the online form, but it could not make a policyholder's circumstances disappear. I prefer to put such costs beside the digital results when speaking to the executive team.

My leadership style is deliberate about ownership. I ask specialists to disagree early, bring the underlying example and say where their own authority ends. I expect programme sponsors to revisit their benefit claims after colleagues have used the system for several weeks. I remain interested in a broader CIO mandate and in a carefully matched board contribution. In either setting, I want my record to be examined through decisions, costs and unresolved limits, rather than through the size of a transformation label.