Chief Human Resources Officer / United States
Decisions that changed what the bank asked of its people
- Colleagues
- 11,600
- Callowmere Financial Corporation, June 2026
- Annual compensation and benefits expense
- US$1.38bn
- Enterprise-wide, 2025; not the HR operating budget
- Critical roles with evidenced successors
- 14 of 17
- Board-designated roles, 2026
Decisions & consequences
01Taking product points out of the branch incentive plan
In 2023 Callowmere paid 2,900 branch colleagues through a points table covering fourteen products, while the Kelverdine branches we were acquiring paid their teams on shared results. Our legacy network logged 312 complaints a year from customers who said they had not requested a product, and 23% of new checking accounts closed within ninety days. After the May 2024 systems conversion we could not run two plans, and whichever plan survived would tell 2,900 people what the combined bank valued.
I proposed extending Kelverdine's team model with changes of our own: a branch pool funded by branch contribution and by accounts still active at ninety days, a smaller individual element based on quality reviews scored outside the branch's reporting line, and removal of that element after any substantiated conduct finding. Our head of consumer banking opposed the proposal in writing. He expected openings to fall by 15–20% and feared losing top producers to competitors. Our chief financial officer was concerned about deposit growth during intense rate competition. Both memos went to the executive committee with my paper.
Our chief executive agreed to a two-quarter pilot in forty branches, compared with forty matched branches on the old plan. Openings in the pilot fell 8%, the share active at ninety days rose fourteen points and complaints fell by two-thirds. The conduct-risk team co-designed the quality measures and signed the final design. We extended the plan across the network from July 2024.
In 2025, the first full year, complaints about unrequested products fell to 41 and ninety-day closures to 9%. New checking accounts fell 9%, but the share still active at ninety days rose from 61% to 78% and average new-account balances rose by about a third. Branch incentive spending fell by US$2.4 million, to US$29.2 million, and voluntary turnover among universal bankers fell from 31% to 24%.
Every customer-facing plan now carries a conduct-risk signature before approval. What I would change is the transition. Eleven of our top-decile producers left within six months. A twelve-month incentive floor costing about US$0.9 million would probably have kept several of them, with their customers. I also left managers' dashboards counting openings for four months, a mixed message I should have caught.
A retention pool built from names rather than salary grades
When Callowmere agreed in February 2023 to acquire Kelverdine Bank, our advisers proposed a US$38 million retention pool for roughly 600 Kelverdine employees above a defined salary grade. Competitors were already calling Kelverdine's commercial relationship managers and credit officers in Knoxville, Chattanooga and Asheville. The deal model assumed 11% attrition in Kelverdine's commercial client book, and much of that book's value sat with a few dozen people who knew the borrowers and the history of their credits.
I argued for a smaller pool aimed at named people. I asked Kelverdine's heads of commercial banking, credit and operations to identify colleagues whose departure would cost client relationships or control knowledge, and to give the reason in a sentence. The list came to 212 people and US$14.6 million, paid in three tranches over twenty-four months and forfeited after a substantiated conduct finding. Kelverdine's chief executive preferred the broader pool, arguing that a grade-based rule was fairer and easier to defend to staff. He was right that it was easier to defend. Our integration director wanted the decision within a week; naming people properly took three.
We gave up the simplicity of a rule and accepted that people left off the list would notice. Each named colleague heard about the award from their own manager rather than a letter. We published the criteria internally without the names and reviewed the list again before the systems conversion, adding fourteen people whose roles had grown.
At twenty-four months, 191 of the 212 remained. Attrition in the acquired commercial client book was 4%, against the 11% in the deal model, and we paid US$13.9 million after forfeitures. The cost of the narrower design showed elsewhere. Nine of about forty next-tier relationship managers who were not on the list left in the first year, and their exit interviews mentioned the list more often than pay.
Naming critical people and recording why has since become the basis of our succession work. If I did it again, I would publish the criteria on the day of the announcement rather than weeks later, because people inferred the list from who was invited to which meetings. I would also create a smaller recognition award for the next tier. It would have cost perhaps US$1.5 million and might have kept several of the nine.
Answering a 68% say-on-pay vote
At our 2024 annual meeting, 68% of votes cast supported our executive pay, compared with 94% the year before. The difference came from one decision. In 2023 the compensation committee had granted one-time integration awards worth US$9.2 million in aggregate to six executives, vesting over three years on continued service. I had recommended time-based vesting because I thought integration milestones would be hard to measure without letting the executives define success for themselves. Proxy advisers and several large holders read the awards as extra pay for work the executives were already paid to do.
The stakes were a second weak vote, strained relationships with our largest shareholders and the committee's own credibility. Our chief executive and two directors wanted to defend the awards and keep the freedom to grant similar ones in future acquisitions. One large holder asked us to convert the existing awards to performance vesting. The committee chair, Lionel Brackwater, and I proposed a third course: honour the awards already granted, which were contractual, commit publicly to no further one-time awards and change the ongoing plan.
That meant giving up flexibility our chief executive valued. It also meant my telling the committee that the time-vesting recommendation had been mine and had been wrong. The committee raised performance shares from 50% to 60% of long-term incentives, measured them on return on tangible common equity relative to eighteen peer banks and on growth in tangible book value, and added a risk modifier attested by the chief risk officer. It extended the clawback policy beyond the restatement recovery that listing standards require, to cover misconduct and significant risk-management failures. The chief executive's 2025 long-term award was held flat rather than raised as peer data suggested.
The chair and I met holders of about 57% of our shares. We rewrote the compensation discussion and analysis around a one-page account of what had changed and why. Support was 89% in 2025 and 93% in 2026.
The committee now reviews any proposed one-time award against a short written note on how shareholders are likely to read it. What I would do differently is plain. In 2023 I should have proposed performance conditions on integration milestones that the audit committee could verify, such as conversion dates, client retention and control findings, and explained them at grant.
Redeployment before reduction in bank operations
Between 2024 and 2026, automation of account maintenance, payment exceptions and loan-servicing tasks removed about 310 roles' worth of work from our operations centres in Roanoke and Knoxville. The 642 colleagues whose work changed had an average tenure of fourteen years, and in both towns the bank is among the larger employers. At the same time our fraud operations team carried an 18% vacancy rate and client service relied on contractors. The finance plan assumed the full US$21 million annual saving from the third quarter of 2025.
I proposed redeployment before reduction: nine months' notice of role changes, paid retraining of eight to twelve weeks for fraud operations, client service and credit administration, and a freeze on external hiring in those functions for roles affected colleagues could fill. Our chief financial officer wanted reductions aligned with each automation release. Our chief operating officer preferred contractors in fraud operations because they could be released if volumes fell. The chief executive accepted my proposal on condition that we reported the cost of delay to the board every quarter.
We gave up about two quarters of savings, roughly US$7.8 million, and some flexibility in fraud staffing. Every affected colleague received a skills assessment and a job preview, and anyone moving to a new role had sixty days to change their mind and take the severance terms instead. Receiving managers were measured on the first-year retention of the people they took on.
By March 2026, 352 colleagues remained in redesigned operations roles, 194 had moved into new internal roles, 41 had retired or left voluntarily and 55 had taken severance, with twenty vacancies left unfilled. First-year retention of redeployed colleagues was 88%. The fraud operations vacancy rate fell from 18% to 4%, and contractor spending in client service fell by US$3.1 million a year. The run-rate saving reached US$19.4 million from the first quarter of 2026, below plan because training and higher grades cost more than expected.
Automation investment cases now include a workforce section and a notice period before approval. What I would change is the training design. We assumed classroom learning would suit everyone, and the first fraud operations cohort passed at 71%, lowest among colleagues who had not studied formally for many years. Paired learning on live queues raised later cohorts to 90%. We should have started there.