Downtime Math for a Two-Person IT Shop: When a Claim Is Worth the Hours
August 30, 2026
A claim is only worth filing if the credit beats the cost of your time. For a two-person IT shop or a solo operator, that sentence is the whole business case, and the arithmetic is less forgiving than the marketing for claim tools suggests. Here is the worked version, with real schedules, so you can decide before the next outage instead of during one.
What an hour of claim work costs you
The claim itself, done on a template with evidence in hand, runs 30 to 90 minutes: assemble incident dates and times, export the relevant log slice, fill the provider's claim format, submit, and log the deadline. Say an hour. If your billing rate is $75 an hour as a consultant, or your loaded cost is $50 an hour as a sysadmin, the claim costs you $50 to $75 in time before anything comes back.
That is the unit cost. It repeats per claim, and the honest small-shop frequency for a single-provider stack is a couple of credit-eligible incidents a year at best, based on how regional outages distribute. The question is never "can I get money back" but "does this month's breach clear the bar".
What the credit actually pays
Credits are a percentage of what you paid for the affected service in the affected region for that month. Not your whole cloud bill: the one service, in the one region. The schedules below are the published ones.
| Provider and service | First credit tier | Credit at tier | Tier above it |
|---|---|---|---|
| AWS S3 | Below 99.9% monthly uptime | 10% of that region's S3 charges | 25% below 99.0%, 100% below 95.0% |
| AWS EC2, multi-AZ | Below 99.99% | 10% of that region's EC2 charges | 25% below 99.0%, 100% below 95.0% |
| AWS RDS Multi-AZ | Below 99.95% | 10% of that region's RDS charges | 25% below 99.0%, 100% below 95.0% |
| Azure VMs across zones | Below 99.99% | 10% of service fees | 25% below 99%, 100% below 95% |
| Google Compute multi-zone | 99.0% to under 99.99% | 10% of that service's regional bill | 25% below 99.0%, 100% below 95.0% |
| DigitalOcean Droplet | Any month below 99.99% | 100% of that Droplet's charge | N/A, single tier |
Now the worked example. A two-person shop runs a $600 monthly AWS bill: $400 EC2, $150 RDS Multi-AZ, $50 S3, all in us-east-1. A regional incident pushes the month's EC2 uptime under 99.99%, so the first tier pays 10% of $400: $40. An hour of paperwork at $60 loaded cost nets negative $20. The same shop with a $4,000 monthly EC2 bill in one region clears $400 at the first tier, and the hour pays back six times.
| Monthly bill (affected service, affected region) | First-tier credit at 10% | Your hour at $60 | Net |
|---|---|---|---|
| $100 | $10 | $60 | -$50 |
| $300 | $30 | $60 | -$30 |
| $600 | $60 | $60 | $0 |
| $1,200 | $120 | $60 | +$60 |
| $4,000 | $400 | $60 | +$340 |
The break-even sits near a $600 monthly spend on the affected service in the affected region. Below that, claims are a hobby. There is one important exception to the whole table: DigitalOcean's Droplet SLA pays 100% of the affected Droplet's charge, so even a $40 Droplet justifies its ten-minute email claim, and a $200 node makes it obvious.
Credits scale with your bill, your claim effort stays flat. The same hour that loses money at $100 a month of exposure pays for itself at $600 and compounds at $4,000.
When the math changes shape
Three situations bend the break-even down. First, severe breaches: if the month lands under 99.0% or 95.0%, the schedule pays 25% or 100% instead of 10%, so even a small bill can justify the hour. Check which tier the incident period actually implies before writing the claim off.
Second, stacked incidents in one claim. AWS expects the dates and times of each incident of non-zero error rates in a single claim for the billing cycle, so a bad month with three qualifying windows is still one hour of work, not three.
Third, your rate is not your only cost. A claim you file also teaches you the evidence trail, the portal, and the deadline tracking, which cuts the next claim to half an hour and makes the marginal claim profitable at a lower bill size. That learning curve is real and it is why shops that file once tend to keep filing.
What a two-person shop should actually do
Set the bar at roughly $500 of monthly spend on a service-region pair and file when an incident crosses it; skip it below. Put an external monitor on the endpoints behind that spend so you know about breaches the week they happen, because the deadline math gives you 60 days on AWS and Azure, and missed windows turn every table above into zeros. The claim guide has the per-provider filing route, and the evidence checklist keeps the hour from stretching into three.
If you would rather not do the watching at all, that is the niche UptimeAudit occupies: detect the breach on your monitored services, draft the claim, track the deadline, and leave you the ten-minute submission. For a shop above the break-even line, the subscription is cheaper than the first claim it catches.