Green only when independent probes agree.
StackWitness watches your services and your dependencies from multiple regions outside your cloud account. A component reads Operational only while independent probes reach consensus. The moment they split, it turns amber, visibly. No self-reported green, ever.
Never a false green: Operational only while independent probes agree. Split quorum turns amber, with glyph and word.
A status page that stays up during your outage.
Public truth pages and customer status pages run on separate infrastructure from the stack you monitor. When you are down, the page still measures, still updates, and still refuses to paint a comfortable green without evidence.
Separate infrastructure
Probes and pages are not hosted inside your account. An outage on your side does not silence the instrument that is measuring it.
Degradation shown honestly
Partial and regional failures surface as degraded, with glyph, shape, and word. We never average a split result into operational.
Measured uptime and history
Uptime labels come from measured windows. When data is insufficient, we say so instead of inventing a percentage.
Independent probes, then consensus, then an honest verdict.
How we measure: multi-region probes outside your network, a quorum requirement before green, and a witnessed-vs-operator split so self-report never pretends to be independent measurement.
Probe from many regions
We hit real endpoints from independent points of presence. A single vantage is not enough to claim operational.
Require consensus
Operational needs agreement. When probes disagree, the quorum device and the status pill both go amber. Never a false green.
Report the honest state
Them vs you: vendor self-report sits beside independent measurement. Attribution pages the right person, with timestamps.
Know if it is them or you, then page the right person.
When checkout fails and the vendor status page still says operational, independent probes already measured reachability from outside. You stop debugging a fine deploy and post an honest update.
Witnessed, not relayed
Dependency truth pages ping real service hosts and also read vendor feeds. The strong signal is unreachability while the vendor claims operational.
A pulsing green dot next to a stale timestamp is a lie. We refuse to show one.
When StackWitness cannot verify, the freshness readout turns unconfirmed. Prior snapshots are not sold as current operational green.
Freshness readout stays signal-colored only while probes still confirm. Status pills keep glyph, shape, and word.
Honest self-stale treatment: when we cannot verify, we say so. No naked green on a dead instrument.
Start free. Upgrade when you need more monitors and a team.
Watch is free forever. Pro is $29/mo with a limited launch price of $20. Team is $79/mo. Enterprise is custom.
Unlimited status-page subscribers on every plan- 5 monitors
- Public status page
- Independent dependency truth
- Email alerts
- Everything in Watch, plus
- 25 monitors
- Private and audience pages
- Faster check intervals
- Slack and webhook alerts
- Everything in Pro, plus
- 100 monitors
- More seats and pages
- Priority support path
- Everything in Team, plus
- Custom monitor volume
- Team roles and access control
- Security questionnaire support
Measurement you do not have to take on faith.
Outside your account
Probes run from independent points of presence outside the monitored cloud account. If your provider is having a bad day, the probes see it, because they are not inside the blast radius.
Witnessed, not relayed
We measure endpoints ourselves and timestamp readings. Vendor status feeds are shown as operator self-report, never re-labeled as independent measurement.
Honest by construction
Green needs a quorum. Stale refuses to look live. We report measurements, never a compliance verdict.
Put a monitor on the dependency that scares you most.
Five monitors free, no credit card. Publish a status page that stays honest, even during your worst day.