Website Monitoring Answers

Sign up for a free monitoring service like Lemwatch, paste your site's URL, and it will ping the page every 5 minutes from multiple regions and alert you by email when it goes down — no credit card required.

Check every 1 to 5 minutes. Use 1-minute checks for revenue-critical pages (checkout, login, API) and 5-minute checks for marketing and content pages where slower detection is acceptable.

99.9% (three nines) is the industry standard for business websites — about 8.7 hours of downtime per year. Revenue-critical SaaS should target 99.95% (4.4 h/year); infrastructure providers commit to 99.99% (52 min/year).

Pick a monitor that probes from at least 5 continents, enable multi-region checks, and require confirmation from 2 regions before alerting — this catches CDN edge failures and regional ISP peering issues that single-region monitors miss.

Most slow websites are caused by one of five things: large unoptimised images, render-blocking JavaScript, a slow origin server, missing CDN caching, or third-party scripts (ads, chat, analytics) that block the main thread.

Use an SSL monitor that connects over TLS, reads the certificate's notAfter field, and emails you 30, 14, 7 and 1 day before expiry. Lemwatch does this automatically for every site you add.

Point a visual-monitoring tool at the URL, set an hourly or daily check, and it screenshots the page, diffs each capture against the last baseline, and alerts when the visual difference crosses a 2–5% threshold.

Send a scheduled HTTP request to the endpoint, assert that the status code is 2xx and the response body contains an expected JSON field (e.g. "status":"ok"), and alert when either check fails twice in a row.

Allowlist your monitoring service's IP range or a shared secret header in a Cloudflare firewall rule, then point the monitor at your public hostname — not the origin IP — so you also validate that Cloudflare itself is serving traffic.

Check the site from three independent spots — a different network, a VPN, and an external checker like Lemwatch's is-down probe. If two report the same failure it's a real outage; if only your network fails, it's local.

5xx server errors (500, 502, 503, 504) always mean an outage. 4xx errors (403, 404, 429) usually don't — the server answered but rejected the request. Connection timeouts and DNS failures also count as outages.

Require two consecutive failed checks from at least two different regions before alerting, add a short re-verify pause (3–5 s) between the failures, and exclude bot-blocked (403/429 from WAF) responses from the failure count.