I've been benchmarking SLAs and service credits across cloud providers and SaaS vendors as part of a larger study on contractual risk. A common finding is that credit formulas are often mathematically flawed, creating the illusion of recourse without meaningful compensation.
The vendor in question uses a typical, but egregious, structure:
* Credit = (Total minutes of downtime in month / Total minutes in month) * Monthly fee.
* "Downtime" only counts if it's a complete platform outage, not degraded performance.
* Credits are capped at 15% of the monthly fee.
The core issue is the denominator. A 4-hour outage in a 30-day month (43,200 minutes) yields a credit of (240 / 43200) * Monthly Fee = 0.56% of your fee. For a $10k/month service, that's $56. The cap is irrelevant at this scale. For the customer, the business impact is orders of magnitude higher.
To fix this in negotiations, propose replacing their formula. The goal is to align the credit with a realistic proportional loss of service value. Here is a more equitable calculation you can put forward as a replacement clause:
```text
Service Credit Calculation:
1. Define "Service Unit" as a consecutive 60-minute period of Downtime.
2. For each Service Unit of Downtime, Customer is entitled to a Credit equal to (1 / 720) * Annual Contract Value.
3. Total Credits for any billing month shall not exceed 50% of the monthly fee.
Example:
- Annual Contract Value: $120,000
- Single Service Unit (1 hour) Credit: (1 / 720) * $120,000 = $166.67
- 4-hour outage Credit: 4 * $166.67 = $666.68
```
This structure is superior because:
* It uses the *annual* value as the baseline, acknowledging that downtime disrupts a full year of operations, not just a month.
* The denominator (720 = 24h * 30d) represents the total hours in a typical billing month, making each hour of downtime a tangible fraction.
* It provides a predictable, non-trivial compensation per hour of outage.
Benchmark data shows that top-tier providers often accept such amendments, especially if you are bringing significant business. The argument is straightforward: their current formula renders the SLA commercially meaningless. Present the alternative as a standard industry practice for mission-critical services.
BenchMark
Your analysis nails the core problem: the denominator is designed to make credits negligible. I call this the "infinite denominator" trick in my own benchmarks.
Your replacement formula snippet got cut off, but the direction is right. I push for a formula based on *business hours* or *expected uptime windows* instead of 24/7/365. For a business-hour service, a 4-hour outage during a 10-hour day is 40% loss of that day's service value.
Also, never let them define "downtime" as only complete outage. That's a huge loophole. Include performance degradation below a defined threshold (e.g., latency > 2s, error rate > 5%). Otherwise, the credit is as useful as a screen door on a submarine.
garbage in, garbage out
Spot on about performance degradation being a loophole. I've seen vendors define "complete outage" so narrowly that a 95% error rate for an hour doesn't trigger anything. It makes the credit clause functionally decorative.
Your business-hour denominator is the right fix. But getting a vendor to accept it often means redefining the entire service's value proposition, which is a much bigger negotiation. Be prepared for pushback that their service is "always available," even if your users aren't.
mod team