The common shortcut — hourly revenue times hours down — measures only lost transactions. Here's the four-layer model that gets infrastructure investment approved.
Why the standard formula undercounts
The common shortcut — hourly revenue times hours down — measures only lost transactions. It misses the cost of the incident response team's time, the customer-support surge handling the fallout, the SLA credits owed to affected clients, and the harder-to-quantify erosion of trust that shows up in the next renewal conversation, not this quarter's numbers.
The four cost layers
- —Direct revenue loss: transactions that don't happen or don't get recovered later
- —Recovery cost: engineering hours, war-room time, vendor escalation fees
- —Contractual cost: SLA penalties, credits, breach-of-contract exposure
- —Reputational cost: real, but should be estimated conservatively — a defensible number is more useful internally than an inflated one nobody believes
A worked example
A mid-market SaaS company with $2M in annual recurring revenue and a four-hour outage: direct loss might be modest, perhaps $3,000 in that window. But add 40 engineering hours at a fully loaded rate, SLA credits to enterprise accounts under contractual uptime guarantees, and a support team fielding three times the normal ticket volume for the following two days — the real number is commonly five to eight times the naive calculation.
Turning this into a budget argument
Infrastructure resilience spending competes against feature work for budget, and it loses when the cost of the status quo isn't quantified. Present the four-layer number, not the naive one, alongside the cost of the redundancy or monitoring investment being proposed. The comparison that actually moves budget decisions is total avoided cost versus investment cost — not "uptime percentage," which executives outside engineering have no intuition for.