For e-commerce, fintech, or any high-transaction business, functional definitions alone are insufficient. A feature can be "working" in a technical sense while silently hemorrhaging revenue. You need financial thresholds as a parallel classification track.
The Dollar-Per-Minute Rule
Mature organizations define severity levels directly in terms of revenue impact. This makes the business cost of an incident immediately legible to everyone — engineers, managers, and executives.
A common structure looks like this:
| Severity | Revenue Impact |
|---|---|
| SEV1 | > $10,000 lost per minute |
| SEV2 | > $1,000 lost per minute |
| SEV3 | < $1,000 lost per minute |
To use this, you need a baseline revenue-per-minute figure. For a business processing $50M/month, that is roughly $1,157/minute. A 10% drop in conversion rate during peak hours might represent $115/minute — a SEV3. A complete checkout outage during a flash sale might represent $50,000/minute — an unambiguous SEV1.
Why Financial Thresholds Override Functional Ones
Consider this scenario:
- Incident A: The "Free User Sign-up" page is completely broken (100% error rate). Zero revenue impact because free users generate no immediate transaction revenue.
- Incident B: The "Enterprise Checkout" flow has elevated latency — 10% of transactions are timing out. Revenue impact: $50,000/minute.
Using a purely functional matrix, Incident A looks like a SEV1 (Total Unavailability, Global Reach). Incident B looks like a SEV2 (Performance Degraded, Major Segment). But the financial matrix correctly identifies Incident B as the SEV1 that demands immediate all-hands response.
The rule: take the higher severity from either the functional matrix or the financial matrix. If the functional matrix says SEV2 but the financial matrix says SEV1, the incident is a SEV1.
Calculating Your Revenue-Per-Minute Baseline
Monthly Revenue / (30 days × 24 hours × 60 minutes) = Revenue per minute
Example: $10,000,000 / 43,200 = ~$231 per minute
This baseline lets you quickly estimate impact during an incident. If checkout is down and you process $231/minute on average, but this is peak hour (3× baseline), you are losing roughly $693/minute — a SEV2 trending toward SEV1 if it persists.
Key Point: Financial thresholds prevent the trap of classifying a high-revenue-impact incident as low severity because it "looks minor" functionally. Always run both tracks and take the higher result.