Uptime monitoring: priced per thing you watch, and one vendor has no plan price at all

Uptime and status monitoring has a natural unit — the number of endpoints you watch — and every vendor in the category prices against it. What differs is how, and the differences are structural rather than cosmetic. Comparing monthly figures across them is close to meaningless without knowing how many things you intend to monitor and how many people need access.

UptimeRobot takes the conventional approach and names its tiers by monitor count: a small tier, a hundred-monitor tier and a two-hundred-monitor tier. Divide each price by its allowance and the cost per monitor falls steadily — the entry tier's rate per monitor is more than two and a half times the top tier's. That is ordinary volume pricing and it means the tier you land in matters more than the vendor you choose, because moving up one rung can lower your unit cost by more than switching provider would.

Cronitor does something our records had not seen before: it publishes no plan price at all. Its paid tier is a formula in two independent variables — a flat monthly rate for each monitor, plus a separate flat monthly rate for each user — charged simultaneously. There is no figure to put in a price column, because the price is a function rather than a number. A team with many endpoints and one engineer and a team with few endpoints and a large on-call rota pay very differently under that model, and neither of them can find their number on the pricing page.

That contrast is the practical heart of the category. Under a bundled model, users are usually free and monitors are the constraint, so a large team watching a handful of services pays little. Under a two-axis model, each additional person on the rota costs the same as several monitors. Work out both counts before comparing anything: the answer genuinely reverses depending on which of the two numbers is larger for you.

The category also contains the tidiest annual discount in our whole record, which is worth a moment. StatusCake's two paid tiers are each priced so that the yearly rate works out at ten months for twelve — and the derivation is visible in the rounding. Both tiers land within a thousandth of a percent of the same figure, and both come out at a fraction over ten months paid, because the vendor computed ten-twelfths and rounded to the nearest penny. That is what a genuinely rule-based discount looks like when you check it: not a round percentage, but a clean fraction with the rounding error showing.

Healthchecks does something else worth crediting, which is printing both annual figures at once — the yearly total and the per-month equivalent, side by side, with the first exactly twelve times the second. Most vendors publish one or the other and leave you to derive the missing one; publishing both removes any doubt about which quantity you are looking at, and lets the arithmetic be checked without clicking anything.

Pingdom sits at the other extreme and shows a failure mode worth recognising on any pricing page. Its markup carries the shape of a price — a currency symbol and a per-month qualifier — with no number between them. The scaffold is served and the value is not. That is more misleading than a blank space, because the eye reads the furniture and supplies a figure that was never there, and it is the reason we record such a page as unread rather than as free or as cheap.

So the sequence for this category is: count your monitors, count the people who need access, and only then look at prices. Compute the cost per monitor at your own count rather than reading the tier names, since the unit rate can differ by more than a factor of two across one vendor's own ladder. And if a vendor prices in two variables at once, accept that no comparison table will ever contain its number and do the multiplication yourself.