A bank's outsourced core banking platform goes down for nine hours on salary day. Compute the direct loss, explain why it understates the damage, and set out the outsourcing controls the bank should have had.

The direct loss is Rs 38 lakh: Rs 18 lakh of lost income and Rs 20 lakh of compensation. That is the small part. If 1% of 30 lakh salary accounts move elsewhere, the bank loses about Rs 9 crore of revenue a year. The bank, not the vendor, answers to customers and the regulator, so it needed a tested recovery target, an exit plan and audit rights; a working two-hour failover would have cut the direct loss to about Rs 6 lakh.

Hire a caterer for a wedding and the caterer fails to turn up: the guests do not blame the caterer, they blame the family that hired it. Suvimal is the family here. Start with what can be counted. Six lakh failed transactions at Rs 3 each is Rs 18 lakh of lost income, and Rs 100 to each of 20,000 customers is Rs 20 lakh, so the direct loss is Rs 38 lakh. Some of the failed payments will be retried later in the day, so the lost income may be a little lower, but the compensation is paid regardless.

Now look at when the failures happened. Salary day loads the morning: people pay rent, card bills and school fees within hours of being paid. Only 100 thousand of the 6 lakh failures fell in the first two hours; the other 500 thousand came after the point where a working backup site should have taken over.

Because a salary account is a relationship, and a failure on the one day customers most need the bank is the failure they remember. If just 1% of the 30 lakh salary accounts move to another bank, that is 30,000 accounts at Rs 3,000 a year, about Rs 9 crore of revenue lost every year, roughly 24 times the direct loss. The 1% is an assumption, not a measurement, and the right figure could be lower or higher; the point is that a small share of a large base outweighs everything on the incident report.

Two further costs sit outside the Rs 38 lakh. The regulator will ask why a critical service had no working backup, and it can restrict the bank's digital onboarding or impose a penalty; the banking regulator's rules on outsourcing IT services set out what it expects, and the current text should be confirmed rather than recalled. And the bank's own staff spend days on complaints, reversals and reconciliation, a cost that rarely reaches the loss event databaseThe record a bank keeps of every operational loss, with its cause, amount and date, used to measure operational risk and to spot repeating failures. at its full size.

Outsourcing moves the work to the vendor; it does not move the accountability. The controls follow from that. Before signing: due diligence on the vendor's resilience, and a record of which other banks share the same platform, since one failure there hits them all. In the contract: a recovery time objectiveThe longest a service may be down before it must be restored, agreed in advance; for payments on a busy day it is usually measured in hours, not a working day. for payments of two hours or less, the right to audit, and an exit plan that says how the bank would move to another provider. In operation: a failover to the backup site tested at live volumes at least once a year, with the bank's staff watching, and an incident playbook that tells customers within the hour what is happening.

Price the control against the loss. A failover that worked within two hours would have cut failures to 100 thousand and the direct loss to about Rs 6.3 lakh, and most of the reputational damage with it. The limitation is cost: a hot backup site that can take full volume at a moment's notice is expensive, and the vendor will charge for it. The board's decision is whether that annual cost is smaller than the loss it prevents, and on these numbers, with Rs 9 crore of revenue at stake, the comparison is not close.