How often should an uptime monitor check your site?

Check interval decides how long an outage goes unnoticed. How to pick between 30 seconds, 1 minute and 5 minutes, and which services deserve the faster setting.

· 6 min read · By the Spot Downtime team

The check interval is how often your monitor requests the URL. It sets a hard limit on how quickly you can learn about an outage: a monitor that checks every 5 minutes can take nearly 5 minutes to notice a problem that started right after the last check. Faster is not always better, though. This post shows how to choose an interval from what an outage costs and how you plan to respond.

What the interval actually controls

Your time to detect a failure is, in the worst case, one full interval plus whatever extra time the monitor needs to be sure the failure is real. Averaged over many outages, you will wait about half an interval before the next check lands on a failure.

  • 5-minute interval: up to 5 minutes before the next check, 2.5 minutes on average.
  • 1-minute interval: up to 1 minute, 30 seconds on average.
  • 30-second interval: up to 30 seconds, 15 seconds on average.

Detection is only the first part. Your total downtime is detection time plus the time for someone to see the alert, open a laptop, find the cause and fix it. If a human takes 20 minutes to react, cutting detection from 5 minutes to 1 saves a fifth of the outage. If your fix is automatic or the on-call person is already at their desk, it can save half.

Interval and your uptime target

Intervals also decide what you can honestly measure. Uptime percentages are computed from checks: if a monitor only samples every 5 minutes, a 3-minute outage may fall between two checks and never appear at all. For a 99.9% target, the allowed downtime is about 43 minutes a month, so a few missed short outages matter little. For 99.99%, the allowance is about 4 minutes a month, and a 5-minute interval can't even resolve a single incident. See what 99.9% uptime really means and the uptime calculator for the arithmetic.

Match the interval to what an outage costs

A practical way to choose is to ask what one minute of downtime costs for this particular endpoint.

Every 30 seconds to 1 minute

  • Checkout, login and payment endpoints, where each failed minute loses orders or locks customers out.
  • APIs that other products depend on, where your outage becomes their outage.
  • Anything covered by an SLA with a tight monthly allowance.
  • Services where recovery is automatic (failover, restarts) and you want to confirm it worked quickly.

Every 5 minutes

  • Marketing sites, blogs and documentation, where a short outage costs little.
  • Internal tools used during working hours.
  • Staging and preview environments.
  • SSL expiry and domain expiry, which change over days, not minutes.

Longer than 5 minutes

  • Content-change and performance trend checks, where you care about direction over hours or days.
  • Heartbeat monitors for jobs that run hourly or daily. These don't poll at all: the interval is the job's own schedule plus a grace period. See heartbeat grace period and interval.

The cost of checking too often

Shorter intervals are not free:

  • Load. At a 30-second interval from three locations, that is a request every ten seconds on average. A lightweight health endpoint handles this easily. An uncached page that runs heavy database queries may not.
  • Rate limits and bot rules. Aggressive checks can trip a WAF or rate limiter, producing false alerts. If a monitor suddenly sees 429 or 403 responses, see 429 Too Many Requests and allow the monitor rather than slowing it.
  • Noise. More checks means more chances to catch a brief network blip. Fast intervals only work well with confirmation logic that makes sure a failure is real before anyone is woken up. See how to reduce false uptime alerts.
  • Plan limits. Fast intervals usually cost more. Spend them where an outage hurts most.

Check a health endpoint, not an expensive page

For high-frequency checks, point the monitor at a cheap endpoint such as /healthz that verifies the app and its critical dependencies, and keep a slower check on the real homepage or user flow.

A sensible setup

You rarely need one interval for everything. A layered approach works well:

  • The most important endpoint (login, checkout, core API) at the fastest interval you can use.
  • Everything else, such as the website, docs and status page, at a relaxed interval.
  • A heartbeat for each scheduled job, timed to that job's own schedule.
  • Slow checks for SSL and domain expiry.

Then send the critical checks to a channel that reaches a person fast, and the rest to a quieter one. A 30-second check that posts into a channel nobody reads gives you no benefit over a 5-minute one.

How this works in Spot Downtime

The fastest interval depends on the plan: 5 minutes on Free, 1 minute on Starter and 30 seconds on Pro. You can set slower intervals on any monitor. Every failure is confirmed from all three check locations (California, Amsterdam and Singapore) before an alert goes out, so a faster interval doesn't mean more pages for single-location blips. See pricing for the plan details and website monitoring for how checks work.

Downtime cost calculatorEstimate what an outage costs you per minute, to decide how much faster detection is worth.

Get new guides by email

Practical posts on monitoring, outages, error fixes and on-call, when they're published. No spam.

Unsubscribe from any email with one click. See our privacy policy.

Keep reading