sla-calculator.exe
Free MSP Tool

SLA calculator: uptime, downtime and the ticket SLA formula

An SLA percentage is a budget. Allowed downtime is the minutes in the period times (1 − SLA ÷ 100), so 99.9% gives you 43 minutes 12 seconds in a 30-day month. For tickets, SLA compliance is the tickets that met the target divided by the tickets measured, times 100, and a 95% target on 240 tickets lets you miss 12.

%
days
minutes
Downtime budget left Within SLA
31m 12s
Uptime this month: 99.972%
Allowed downtime
Per day1m 26.4s
Per week10m 4.8s
Per month (30 days)43m 12s
Per year (365 days)8h 45m 36s
43,200 min × (1 − 99.9 ÷ 100) = 43.2 min allowed

The SLA formulas

What you wantFormula
Uptime percentage(minutes in the period − downtime) ÷ minutes in the period × 100
Allowed downtimeminutes in the period × (1 − SLA ÷ 100)
Ticket SLA compliancetickets that met the target ÷ tickets measured × 100
Misses you can affordtickets measured × (1 − target ÷ 100), rounded down

The arithmetic is the easy part. The arguments happen over what goes into it. For uptime, downtime means minutes the service couldn't be used, and some SLAs leave scheduled maintenance out of the count. For tickets, "measured" means the tickets under that SLA that closed in the period, and the clock may not run the whole time: in HaloPSA, the With User status puts a ticket on hold by default and pauses the SLA timer, so time spent waiting on the user doesn't count against you. The Ticket SLA tab works for a response-time SLA too: count the tickets that missed the response target instead.

Downtime by SLA level

Here's what each common uptime SLA allows for a service that has to be up around the clock.

SLAPer dayPer weekPer 30-day monthPer 365-day year
99%14m 24s1h 40m 48s7h 12m3d 15h 36m
99.5%7m 12s50m 24s3h 36m1d 19h 48m
99.9%1m 26.4s10m 4.8s43m 12s8h 45m 36s
99.95%43.2s5m 2.4s21m 36s4h 22m 48s
99.99%8.6s1m 0.5s4m 19.2s52m 33.6s
A month's downtime budget, laid on one working day
A month's downtime budget, laid on one working dayOne outage starting at 9:00 am. At 99%, 7 hours 12 minutes, until 4:12 pm. At 99.5%, 3 hours 36 minutes, until 12:36 pm. At 99.9%, 43 minutes 12 seconds, until 9:43 am. At 99.95%, 21 minutes 36 seconds, until 9:22 am. At 99.99%, 4 minutes 19 seconds, until 9:04 am.9:0011:0013:0015:0017:00to 4:12 pmto 12:36 pmto 9:43 amto 9:22 amto 9:04 am
Allowed downtime in a 30-day month, spent as one outage starting at 9:00 am.

Two uptime calculators can give different answers for the same SLA, and the month is why. Take a month as a twelfth of a 365.24-day year instead of 30 days, and 99.9% allows 43 minutes 50 seconds. Set the days to the month your contract measures.

How Microsoft 365's SLA counts downtime

Microsoft measures each calendar month, and it doesn't just count how long an outage lasted. It counts user-minutes: each incident's minutes times the number of users it hit, against every user's minutes in the month. Say a 300-user client loses Exchange Online for 2 hours, but only 50 users are affected. That's 6,000 downtime user-minutes out of 12,960,000 in a 30-day month, or 99.954% uptime. That's above 99.9%, so no credit. The Exchange Online credit is 25% below 99.9%, 50% below 99% and 100% below 95%, and credits are your only remedy. Your client felt that outage a lot more than 0.046% of their month, so log it whatever the SLA says.

What resolution target can L1 tickets actually hit?

Uptime math is fixed. Ticket targets are a judgment call, and the easy way to make it is to look at how fast a typical ticket closes. Rallied's own numbers show why that's a trap. Across 2,099 tickets Rallied worked in October 2026, half closed within 5.3 hours of being opened, and a quarter took longer than 46 hours. Set a resolution target at the median and you miss half your tickets by design. Set it at the 75th percentile and you still miss a quarter. A 95% target has to sit beyond both.

Hours from open to close: median and 75th percentile, by ticket type
Hours from open to close: median and 75th percentile, by ticket typePassword reset or lockout: median 2.3 hours, 75th percentile 17.0 hours. User offboarding: median 2.6 hours, 75th percentile 20.7 hours. MFA or sign-in issue: median 2.8 hours, 75th percentile 26.6 hours. User onboarding: median 3.2 hours, 75th percentile 73.1 hours. Permissions or access: median 4.9 hours, 75th percentile 27.2 hours. Email or mailbox: median 5.0 hours, 75th percentile 41.4 hours. All tickets: median 5.3 hours, 75th percentile 46.3 hours.0 h20 h40 h60 h80 h
Filled dot: median. Open dot: 75th percentile. 2,099 tickets from people that Rallied worked in October 2026, each type 75 tickets or more. Calendar time from open to close, nights and waiting on the user included. Ticket types are labelled by a model, so treat them as estimates.

A paused SLA clock leaves out the waiting on the user, so read these hours as the most your clock would show. The shape holds anyway, and it's why we'd set resolution targets per ticket type, well past each one's 75th percentile. Onboarding's is 73.1 hours, more than four times the 17.0 hours for password resets. As one MSP put it on a call with us, "A customer is waiting for an hour for one of the L1 guys to pick up the ticket and then it gets resolved in 30 seconds." That wait is the tail, and it's what breaks the SLA. We go deeper on where ticket time goes in MSP ticket resolution time benchmarks and how to cut first response time, and the MSP ROI calculator puts a cost on the same tickets.

Close L1 tickets inside your SLA

Rallied is an AI agent that works password resets, MFA issues, onboarding and offboarding inside ConnectWise, HaloPSA or Autotask. It runs the fix in Microsoft 365 and your RMM, then writes the notes back to the ticket. It's resolving tickets the same week you connect it.