Every monitoring tool can tell you your site went down. That has been a solved problem for two decades. The problem nobody has solved is the one that actually does damage: being told your site is down when it isn't, often enough that you quietly stop believing the tool at all.
Tallwatch is built for that second problem.
The failure mode that matters is not technical
It is human. A monitor that cries wolf does not fail by crashing. It fails by training you to ignore it.
It is worse than most teams admit out loud. Survey after survey of on-call engineers tells the same story: most alerts get dismissed without investigation, most are judged false positives, and a typical rotation fields far more pages in a week than anyone acts on. Ask any engineer who carries a pager and they will recognize it instantly. Every wasted page is a small deposit into the same account, and the withdrawal comes on the night a real outage looks exactly like the noise you have been swiping away for months.
So "can it detect a failure" is the wrong question to ask a monitoring tool. They all can. The right question is narrower and harder: when it wakes you at 2am, will it be right. Everything in Tallwatch bends toward making the answer yes.
How it decides something is actually down
A check from a single location measures two things and reports them as one: whether your site is up, and whether the path between that one machine and your site is healthy. They usually agree. When they don't, a bad thirty seconds of routing somewhere in the middle gets filed as your downtime, and your phone rings.
Tallwatch checks each target from several regions and treats every result as a vote rather than a verdict. An incident opens only when a configurable majority of regions agree the target failed inside a short window. A single region having a rough moment is outvoted by the rest, and you sleep through it.
This is multi-region consensus, and it sits on every plan, including the free one, because consensus is the product, not a feature to upsell.
What you get
Tallwatch is the whole pager stack, not a checker bolted to an email:
- Monitoring for any website or API over HTTP and HTTPS, watching the response status from every region and recording latency alongside it.
- Alerts in the seven places teams already live: email, Slack, Discord, Microsoft Teams, Telegram, PagerDuty, and signed webhooks.
- Webhooks you template yourself, so an incident lands in whatever internal tool you have wired up.
- On-call rotations and escalation that keep climbing until a human acknowledges.
- Status pages on your own domain, so customers answer "is it down" without opening a ticket.
- Maintenance windows, so a planned deploy does not page the team at midnight.
Who it is for
People who have been burned by a noisy pager and decided they are done. Teams that outgrew a basic checker. Founders running a side project that quietly pays a bill. The engineers who actually carry the rotation and have learned to care more about signal than about another dashboard.
Where it really is
Pre-launch, and honest about it. There are rough edges, and I would rather you hear that from me. What is not rough is the free tier: multi-region consensus across three regions, on-call and escalation, and a status page on your own domain, with no card and no trial clock.
Point it at something you would hate to see go down, and judge it the only way that counts. Wait for the next time one corner of the internet has a bad night, and see whether your phone stays quiet.