Calling it “minor” costs you 15h 01m
Everyone knows critical incidents get fixed first. Nobody talks about the price of the other label. At the 90th percentile a critical incident has been open 8h 06m. A minor one has been open 23h 07m.
That is 2.9x as long, and 15h 01m of extra downtime, for incidents nobody classified as urgent. One in ten “minor” incidents ran most of a day.
Difficulty is not what separates those two numbers — the medians are close (1h 25m against 2h 02m). Attention is. “Minor” is what an incident gets called when nobody will be paged for it, and then it sits.
| Severity | Incidents | Median | 90th percentile |
|---|---|---|---|
| critical | 152 | 1h 25m | 8h 06m |
| major | 505 | 1h 49m | 12h 20m |
| minor | 1,165 | 2h 02m | 23h 07m |
| Unclassified | 495 | 1h 01m | 13h |
Severity is the provider’s own label, not our reclassification. “Unclassified” is the API’s none impact, used for informational posts and for incidents a provider never graded.
Ten hours of silence, at the worst possible moment
Most updates arrive 46m after the last one. At the 90th percentile the wait is 9h 52m — 12.9x as long.
That is the widest gap in the dataset, wider than time-to-resolve itself (10.2x). Communication falls apart faster than the incidents do, and it falls apart on the long ones: a customer stuck in the worst outage of the year is the one hearing nothing. Writing updates is the first job dropped when the incident is still burning.
8% of incidents got a single post and no follow-up at all — announced, then nothing until “resolved”.
A third of outages end without anyone saying what broke
917 of 2,544 incidents — 36% — closed with no cause given anywhere: no identified status, and nothing in the update text either. Your service was down, it came back, and that was the whole explanation.
When a cause did land it landed fast, at a median of 17m — then a tail out to 2h 12m. So diagnosis is rarely the slow part. Noticing is.
Counting only incidents that never reached identified would give 54%. A third of those explained the cause in prose without moving the status, so we do not count them. 36% is the figure that survives being checked.
You break your own service 4.8x as often on a weekday
Incidents start on weekdays at 4.8x the weekend rate: 469.4 a day against 98.5. Weekend traffic does not drop by anything close to that, so this is not load. It is deploys, migrations and config changes — which means most outages are self-inflicted and land in working hours, when somebody is around to catch them quickly. The ones that run for a day are the ones nobody caught.
| Day started (UTC) | Incidents | |
|---|---|---|
| Mon | 476 | |
| Tue | 475 | |
| Wed | 457 | |
| Thu | 527 | |
| Fri | 412 | |
| Sat | 107 | |
| Sun | 90 |
How long incidents actually run
Of the 2,317 incidents with both a start and a resolution timestamp:
| Time to resolve | Incidents | Share |
|---|---|---|
| Under 30 minutes | 515 | 22% |
| 30 to 60 minutes | 306 | 13% |
| 1 to 4 hours | 837 | 36% |
| 4 to 12 hours | 358 | 15% |
| Over 12 hours | 301 | 13% |
Roughly a third are dealt with inside the hour. The last 13% run past twelve hours, and those are where the communication tail above lives.
Method
Every provider here publishes its incident history through the Statuspage v2 API. We read it, kept every incident that started between 2025-09-07 and 2026-09-07, dropped planned maintenance, and measured from each incident’s own timestamps. 2,544 incidents, 77 providers, all of it public and all of it re-checkable.
No NoDisrupt customer data is in here. A number you cannot verify yourself is not worth quoting, and our customers’ outages are theirs, not our marketing material.
Two things we deliberately do not claim. There is no incidents-per-month figure: the API caps at 50 incidents per provider and 33 of 77 hit that cap, so a rate would understate the busiest providers. And the sample is providers who publish a status page at all — the transparent end of the market, which makes the 36% with no stated cause more striking, not less.
The 77 providers
- bitbucket.status.atlassian.com
- status.1password.com
- status.airtable.com
- status.amplitude.com
- status.appcenter.ms
- status.asana.com
- status.bitrise.io
- status.box.com
- status.braze.com
- status.browserstack.com
- status.bugsnag.com
- status.buildkite.com
- status.circleci.com
- status.cloudbees.com
- status.cloudinary.com
- status.coinbase.com
- status.confluent.cloud
- status.contentful.com
- status.crates.io
- status.customer.io
- status.datadoghq.com
- status.digitalocean.com
- status.discord.com
- status.doppler.com
- status.dropbox.com
- status.duo.com
- status.elastic.co
- status.epicgames.com
- status.expo.dev
- status.figma.com
- status.fly.io
- status.gocardless.com
- status.grafana.com
- status.hashicorp.com
- status.hubspot.com
- status.jfrog.io
- status.kinsta.com
- status.klaviyo.com
- status.launchdarkly.com
- status.linode.com
- status.logdna.com
- status.mailgun.com
- status.miro.com
- status.mixpanel.com
- status.mongodb.com
- status.netlify.com
- status.newrelic.com
- status.npmjs.org
- status.onesignal.com
- status.opsgenie.com
- status.optimizely.com
- status.percy.io
- status.plaid.com
- status.prismic.io
- status.pulumi.com
- status.pusher.com
- status.render.com
- status.rollbar.com
- status.rubygems.org
- status.sanity.io
- status.saucelabs.com
- status.segment.com
- status.sendgrid.com
- status.sentry.io
- status.snowflake.com
- status.snyk.io
- status.sumologic.com
- status.supabase.com
- status.testrail.com
- status.twilio.com
- status.twitch.tv
- status.vercel.com
- status.wise.com
- status.zapier.com
- status.zoom.us
- www.cloudflarestatus.com
- www.githubstatus.com
Questions
Where does this data come from?
The published incident history of 77 software providers, read from the Statuspage v2 API. Public data on a documented endpoint, so every figure can be checked. None of it comes from NoDisrupt customers.
What counts as an incident?
Anything the provider posted to its own status page and later resolved, at any severity. Planned maintenance is excluded. Severity is the provider's own label.
How would NoDisrupt have changed these numbers?
Detection and escalation are the parts that decide a long incident, and both happen before anyone starts fixing anything. Checks every 30 seconds instead of every five minutes, an escalation policy that keeps paging until somebody acknowledges, and a status page fed by the same checks so it updates without anyone remembering to. None of that makes an outage easier to fix. All of it shortens the tail.
Can I get the next edition?
We run this annually. The free plan includes ten monitors and needs no card, and we will send the next edition to anyone on it.