MTTD (Mean Time To Detect)
MTTD is the average time from when an incident begins to when your monitoring notices it.
Mean Time To Detect is the average gap between the moment a failure actually starts and the moment your monitoring flags it. It is the first phase of the incident lifecycle and the one tooling controls most directly — everything else (acknowledging, diagnosing, fixing) can only begin once detection has happened.
Why it matters: detection lag is pure, avoidable downtime. Every second between failure and detection is time customers are affected while no one is even aware. Because MTTD sits at the front of the chain, it adds directly to MTTR and therefore to total downtime; reducing it is often the single highest-leverage reliability improvement a team can make.
A common misconception is that detection is effectively instant. With low-frequency polling it is not: if you check every five minutes, an outage that starts just after a check goes unnoticed for almost the full interval. That interval is added to every incident before anyone responds. Frequent checks — down to one second — collapse this window, and confirming failures from multiple regions keeps detection both fast and trustworthy.
MTTD also depends on covering the right signals. A service can be 'up' at the network level while a critical user flow is broken; detecting that needs checks that assert on the actual response (status code, body content, latency), not just a successful connection.
MTTD = Total time-to-detect across incidents ÷ Number of incidents
If detection took 6, 2, and 4 minutes across three incidents, MTTD = 12 ÷ 3 = 4 minutes. A 60-second check interval caps added detection lag near one minute; 1-second checks cut it to roughly one second.
Frequently asked questions
- How is MTTD different from MTTR?
- MTTD measures only the detection phase — failure start to first alert. MTTR covers the whole incident through to recovery. MTTD is a component of MTTR, so improving detection directly lowers recovery time.
- How do I lower MTTD?
- Check more frequently, monitor from multiple external regions, and assert on the actual response rather than a bare connection. Higher check frequency shrinks the worst-case detection window to roughly one interval.
- Does check interval set my MTTD?
- It sets the ceiling. With a 5-minute interval, an outage can go undetected for almost 5 minutes; with 1-second checks, the worst case is about a second. Interval is the dominant lever on detection lag.
Start monitoring in under a minute
Free forever, no credit card. Upgrade when you need finer intervals.