The Recovery Target That Outran the Budget
A mid-sized insurance claims processor has an RTO of 4 hours for its claims intake system, set two years ago by the board after a competitor's outage made headlines. The current architecture is a single primary data center with nightly tape backups trucked offsite; the last full restore test took 31 hours and required manual re-keying of 6 hours of transactions. The CIO wants to fund an active-active second site at roughly 2.4x the current infrastructure budget. The CFO has offered a warm standby at 1.3x, which the IT team estimates yields a realistic recovery of 9 hours. Meanwhile, the claims VP reports that during a 6-hour outage last quarter, only 14% of claims were time-sensitive; the rest could be batched. No formal BIA has been performed since the RTO was set. The board meets in six weeks.
1. What is the central analytical error in how this decision is being framed, and what should happen before the board votes?
Read the worked answer
The error is treating a board-declared RTO as a fixed requirement rather than a derived output of business impact analysis. The 4-hour target was set by benchmarking against a competitor, not by quantifying the cost of downtime for this organization's actual claim mix. That is an aspiration masquerading as a requirement.
Before the board votes, the program should run a focused BIA that measures downtime impact in financial and operational terms, segmented by claim type and time sensitivity. The claims VP's data point is the key input: if only 14% of claims are time-sensitive, the true impact curve may be steep for a narrow slice and flat for the rest. That suggests a tiered strategy - protect the time-sensitive intake path to a tight RTO, accept longer recovery for batch work - rather than a single system-wide target.
The board should then choose among options against a cost-of-downtime curve, not against a competitor's headline. The recommendation should present the 4-hour, 9-hour, and tiered options with their costs and residual risk, and explicitly state what the board is accepting if it funds the cheaper option.
2. The CFO argues that because the 9-hour option is cheaper, it is obviously the better value. How would you test that claim?
Read the worked answer
Value is not the same as lowest cost; the CFO's framing omits the cost of the risk being retained. The test is to convert the 9-hour gap into expected loss.
Steps: (1) Estimate the financial impact per hour of outage for the time-sensitive claim segment, using the BIA. (2) Estimate annual likelihood of an outage exceeding 4 hours from historical incident data, facility risk, and threat analysis. (3) Multiply impact by likelihood to get expected annual loss attributable to the 5-hour gap. (4) Compare that expected loss plus the 1.3x infrastructure cost against the 2.4x active-active cost.
If expected annual loss from the gap is small relative to the 1.1x budget difference, the CFO is right and the board should document the accepted risk. If it is large, the cheap option is the expensive one. Either way the decision becomes explicit and auditable rather than a cost comparison. Note also that the 9-hour estimate is unvalidated - it should be treated as a planning assumption until a restore test confirms it.