Free Uptime Monitoring: How to Spend the Allowance

Every free uptime monitoring plan is an allowance, not a gift. The vendor has decided how many checks per hour it will run for you, from how many places, kept for how long, and delivered down how many channels. Spend that allowance deliberately and a free tier covers a small production estate properly.

What the word free is buying

Free tiers exist because a monitoring check is cheap to run and expensive to trust. Vendors give away the running and charge for the trusting: faster intervals, more vantage points, longer history and a status page you can put your own domain on.

That framing makes the comparison easier. Stop asking which plan has the most monitors. Start asking how many checks per hour the plan runs, from where, and what happens to the evidence afterwards.

Three things are almost never free anywhere, and they are worth knowing up front. Paging by phone or SMS, history beyond a month or two, and a status page on your own domain. If any of those is a requirement, free uptime monitoring is a starting point rather than a destination.

The first budget line: checks per hour

Multiply monitors by checks per hour before you compare anything else. Fifty monitors at a five-minute interval runs 600 checks an hour. Ten monitors at a one-minute interval runs 600 as well, from a much smaller list.

The second number is the one that decides how quickly you find out. An interval of five minutes means a failure can be nearly five minutes old before the first check even notices it, and confirmation adds more on top.

We tested both shapes against the same deliberately failed endpoint. The one-minute plan raised a confirmed incident while the five-minute plan was still waiting for its next scheduled run. Neither tool was wrong. The allowance had already decided the outcome.

So the practical question is not how many things you can watch. It is which handful of endpoints genuinely need a fast answer, and whether the plan lets you point your whole allowance at them.

There is a second multiplication worth doing. If a check runs from two locations, most vendors count that as two checks against your allowance, and a few count it as one. Read which model applies before you assume a plan gives you redundancy for free. The full set of levers, including the ones that only appear on paid tiers, is laid out in uptime monitoring tools.

The second budget line: how many places have to agree

A check that runs from one location cannot tell a broken service apart from a broken path to it. That distinction is the difference between an incident and a blip, and free tiers handle it very differently.

Some free plans run each check from a single location. Some rotate locations without requiring agreement. As of September 2026, the free plan here runs each check from up to two regions and requires both to agree before a monitor is marked down. Six regions exist in total, and the paid tiers widen how many of them a single check can use.

curl -X POST https://uptimeroom.com/api/v1/monitors \
  -H "Authorization: Bearer urm_live_..." \
  -H "Content-Type: application/json" \
  -d '{
    "name": "public site",
    "type": "http",
    "target": "https://example.com/",
    "intervalSec": 60,
    "regions": ["london", "bangalore"],
    "expectedStatus": [200],
    "alertAfterCycles": 2
  }'

Region choice matters as much as region count. Two probes on the same continent will agree with each other about a transatlantic routing problem, which is agreement without independence. Pick locations that sit on genuinely different paths to your origin, and if your edge filters by source address, the probe IPs page lists what to allow.

Ask each candidate what happens when its own probe network has a bad minute. A vendor that counts its own network errors as your downtime will quietly ruin your availability figures, and the reasons that matters are unpacked in five nines uptime.

What free tiers quietly cut

The headline number is monitors. The cuts are everywhere else, and they show up in week three rather than week one.

History. A month of retention is typical on free plans. That is enough to answer what happened last Tuesday and not enough to argue about last quarter.

Seats. Most free tiers are single-login. Adding a colleague is frequently the reason teams move to paid, not the monitor count.

Channels. Email and browser push are nearly always included. Chat integrations, webhooks, SMS and voice usually are not, and SMS is metered even on paid plans.

Slow-response alerting. Alerting on a service that is up but degraded tends to sit behind the paywall. Availability is binary and cheap; latency thresholds are not.

The status page. Free status pages exist, on someone else's subdomain, with someone else's branding on them.

There is one more cut that nobody advertises. Support on a free plan is a community forum, so a question about a monitor behaving strangely gets answered by another user or not at all. That is fine until the strange behaviour is happening during an outage.

None of that is unreasonable. It is worth knowing before you build a process around a feature you do not have. What noise does to a team that cannot tune its alerts is the subject of alert fatigue.

The free tiers side by side

Every figure below was read on that vendor's own pricing page on 23 September 2026.

Tool	Free allowance	Interval	Notable limit
UptimeRobot	50 monitors	5 minutes	Detection is bounded by the 5-minute schedule
StatusCake	10 uptime monitors	5 minutes	Free forever, with paid tiers for 1-minute tests
Better Stack	10 monitors and heartbeats	Not stated on the page	One status page
Healthchecks.io	20 jobs	Job-driven	100 log entries per job, inbound pings only
Uptime Room	10 monitors	1 minute	Up to 2 regions per check, 1 month of history
Uptime Kuma	Unmetered by licence	Your choice	You run it, and it checks from one place

HetrixTools also advertises a free uptime monitoring tier, but its site refused a direct read on 23 September 2026, so no figure from it is quoted here. Check the page yourself before relying on any number you see repeated in a listicle.

Healthchecks.io is in a different category from the rest of that table on purpose. It watches jobs that call in rather than endpoints it calls out to, which is the only way to catch a scheduled task that silently stopped running. The mechanism is explained in heartbeat monitoring.

Self-hosting is a different kind of free

Uptime Kuma is an MIT-licensed self-hosted monitoring tool, and on licence terms it is genuinely unmetered. The cost moves rather than disappearing.

You pay in a host, in patching, and in a structural blind spot. Checks run from wherever the container runs, so one network path produces every verdict. Put it on the same provider as the service it watches and a regional incident removes both at once.

Budget the maintenance honestly. A self-hosted checker needs a database, backups, upgrades and somebody to notice when the container stops. Teams that skip those discover the gap when a disk fills and the dashboard has been green and static for a fortnight.

We prefer self-hosting for internal services behind a firewall, where an external probe cannot reach anyway. For anything public, the vantage point is the product, and giving it up to save nine dollars is a poor trade.

When does free uptime monitoring stop being enough?

Four signals, and most teams hit the first one before the others.

1. A second person needs a login, and the plan allows one.
2. You need an alert in chat or on a phone rather than in a mailbox.
3. Someone asks what your availability was last quarter and the history window is a month.
4. Customers start asking for a status page on your own domain.

The order matters. Teams almost always hit the seat limit long before they hit the monitor limit, because monitoring is the first thing a second engineer wants access to. Plan for that rather than being surprised by it.

None of those is about monitor counts, which is why teams are surprised when they arrive. Paid entry tiers in this market sit around $9 to $25 a month as of September 2026. Our own Individual plan is $9, with 30-second checks and a year of history. The full set of limits is on the pricing page.

Common mistakes

Buying on monitor count. The allowance is checks per hour, and a big monitor count at a slow interval is a smaller allowance than it looks.

Watching only the homepage. It is usually cached and static, so it is the last thing to break. Spend a free slot on the login or checkout endpoint instead.

Leaving alerts in a mailbox. An alert that arrives where nobody looks at 2am is documentation, not detection.

Assuming a listicle's numbers are current. Free tiers change more often than any other part of a pricing page. Open the vendor's own page before you commit, and note the date you read it.

Skipping certificate and domain expiry. Both are free to check and both take a site down completely. An expired certificate is the cheapest outage you will ever have and the most embarrassing.

Free website uptime monitoring fails the people who spread the allowance evenly and then wonder why the important thing went unnoticed. Used deliberately, free uptime monitoring covers a small estate properly. Pick the plan whose interval and confirmation behaviour you trust, point it at the two or three URLs your revenue depends on, and note the date you compared the tiers. When you outgrow yours, the API reference turns recreating the list into a five-minute job.