Outage confirmation
Fewer false positive uptime alerts: two providers must agree
Pingtier opens as a free beta soon. Sign-up is not open yet.
A false Down alert costs you twice. Someone stops work to check a site that is fine. Next time, the team trusts the alert a little less. Pingtier is built to send a Down alert only when the failure is confirmed.
Why a single check is not enough
One failed request can come from the network path, the probe's own provider or a short blip. It does not prove that the site is down for your client's visitors. A monitor that alerts on one failed check from one place will sometimes alert on nothing.
How Pingtier confirms an outage
Pingtier runs 8 IPv4 probes on 4 different providers. Four probes are in Europe (Lithuania, Germany, Poland and the Netherlands), two in the US (east and west) and two in Asia (Singapore and Tokyo).
- One probe runs the normal check at your interval.
- If that check fails, Pingtier starts a confirmation round. It starts about 15 seconds after the failing check began.
- Up to 3 other probes check the site again. Each one runs on a different provider, and Pingtier prefers providers other than the one that saw the failure.
- The monitor goes Down only when at least 2 probes on different providers report a failure.
- If 2 probes report success instead, the monitor stays Up and Pingtier records a short failure. You get no alert.
When the round confirms Down, Pingtier sends the alert to Slack and email at once. It sends a recovery alert when at least 2 probes report success again.
A failure means the site answered with an unexpected HTTP status, the keyword was missing, the request timed out after connecting, the TLS certificate failed validation, the connection was refused or reset, or the domain does not exist (NXDOMAIN). A problem on the probe itself, such as its own resolver being unreachable, does not count as a vote. An HTTP 429 (rate limited) response counts as Up, with a note.
Choose the regions for each monitor
Each monitor checks from the regions you select: EU, US and Asia. All three are on by default. Each region has probes on at least 2 providers, so a monitor that uses only one region can still confirm an outage.
Some sites answer only visitors from one region. If a site blocks traffic from outside Europe, probes in the US and Asia will fail, and Pingtier can report a Down that your client's visitors do not see. Unselect US and Asia for that monitor. The incident shows the result per region, so you can see this cause.
Blocked is not Down
Bot protection sometimes answers a monitoring request with a challenge page instead of the site. Pingtier recognises these challenge markers:
cf-mitigated: challenge(Cloudflare)x-amzn-waf-action: challengeorcaptcha(AWS WAF)
When the challenge answers the probes, Pingtier sets the status Blocked, not Down. It does not open a Down incident and does not page Slack. It sends one email with the probe IP addresses and the User-Agent to allow, and it shows a Blocked issue on your Today list. A monitor gets at most one Blocked email in 24 hours. The time spent Blocked counts as unknown in uptime, not as downtime. The Blocked issue closes by itself when a check succeeds again.
A plain HTTP 403 without a challenge marker still counts as a failure. Pingtier cannot tell it apart from a real error page.
If a site is Down and the recovery checks meet a challenge page, the incident stays open with a "checks blocked" note. A challenge page does not prove that the site works.
How to allow the Pingtier probes
How other monitors describe confirmation
These are statements from each vendor's public pages, checked on 2026-11-08. We make no claim about their false alarm rates.
| Tool | What its documentation says | Bot challenges |
|---|---|---|
| Pingdom | A failed test marks the check Unconfirmed Down. A probe in a different location retests, and the check is Down only if that retest also fails. (source) | No detection feature documented. The docs advise allowlisting the probe IPs. |
| UptimeRobot | A failed check is re-checked up to 3 times within seconds, each time from a different data center in the same region. (source) | No built-in detection documented. The help centre advises allowlisting its IPs. |
| StatusCake | 0 to 3 confirmation tests, default 2, run from other servers in the same monitoring region. (source) | No detection documented. The docs advise allowlisting its IPs. |
| Better Stack | A marketing page says each HTTP incident is verified from at least 3 locations. (source) | No detection documented. The docs advise allowlisting by User-Agent. |
Compare in detail: Pingdom alternative, UptimeRobot alternative.
Try it on your client sites
The free plan covers 3 clients with 5 monitors each, checked every 3 minutes. Free during beta. Then EUR 4 per client a month, excl. VAT. See pricing or read about website monitoring for agencies.
FAQ
Does Pingtier promise zero false alarms? No. Two-provider confirmation removes alerts caused by one probe or one network. A site that blocks some regions, or answers probes with a plain 403, can still produce an alert that visitors do not see.
What happens if only one probe sees a bot challenge? One challenged probe is inconclusive. It cannot make the monitor Down or Blocked on its own.
What if fewer than 2 probes can check a monitor? Pingtier cannot confirm a change with fewer than 2 eligible probes, so the status stays as it was and the monitor is marked as not observed.
Does Blocked time count as downtime? No. Time in Blocked counts as unknown and is left out of the uptime calculation.