Planning for the hardest queue, night and outage

Most of a credit union's busiest days are predictable. Mortgage maturities are known years ahead, tax-season questions arrive in March and April, and card replacements follow a fraud alert within hours. I build staffing forecasts from those calendars, by half-hour and by language, rather than from last year's average volume. Where demand can be moved, I move it: booking renewal conversations months before maturity spreads the work across quieter weeks. Where it cannot, I pay for the capacity in advance and show the board the cost. I report answer times and abandonment for each queue separately, because a combined figure can look healthy while one group of members waits far longer. I also keep a reserve plan for the day a fault produces calls nobody forecast.

A member who chooses French should not wait longer, receive a shorter explanation or be asked to repeat a fraud report in English. At Brayloche I treat that as an operating standard, measured weekly, not a statement of intent. It shapes where we recruit, because bilingual advisers are easier to find in the northeast of the province than in Moncton, and how we schedule, because the French queue peaks at different hours. It affects procedures too: every branch outage instruction, fraud script and renewal letter exists in both languages before it is used. The cost is real. I pay a language premium, fund translation and sometimes accept slower English answer times during training periods. I explain those trade-offs to the board rather than describing bilingual service as free.

Fraudsters work weekends and holidays, and an automated block cannot ask a member whether a transfer is genuine. Our overnight desk of seven analysts reviews alerts from card and online-banking monitoring and telephones members when a payment looks wrong. The design questions are practical: which alerts justify a call at two in the morning, which holds an analyst may release without a manager, and how a member who does not answer regains access before work. I review false positives as closely as losses, because a blocked card at a pharmacy is also a failure. I work with our risk and compliance colleagues on rules and reporting; they own the fraud risk appetite, and my teams operate within it. Since 2025 the same desk also covers two smaller credit unions under a service agreement.

Storms on the coast are not rare events for us, and our branches serve communities where the nearest alternative may be forty minutes away. I plan continuity around what a branch must still do without its systems: verify identity, pay limited cash, accept deposits for later processing and record every transaction for reconciliation. Those limits now sit in a board-approved procedure, written in both languages and rehearsed twice a year, instead of being decided by telephone on the morning of an outage. I also decide where resilience investment goes. Generators and satellite links protect seven branches chosen for their isolation and the members who depend on them, not all twenty-one. I report that choice to the risk committee as accepted exposure, with the branches it leaves unprotected named.

After the 2016 merger we had branches a few minutes apart in some towns and none for long distances elsewhere. Reducing the network from 29 to 21 was partly a cost decision, but the useful question was what each community still needed in person: cash for fishing crews and farm suppliers, help for members uncomfortable with online banking, and a place to sign mortgage documents. Where we left a village, we kept a cash machine and an adviser two days a week in four places. I track how far members travel, how often they visit and whether complaints come from people we did not reach. I would rather revisit a decision openly than defend a closure because it appeared in an earlier plan. Branch managers see those effects first, so their monthly notes come to me directly.