What a $20/month site-monitoring stack actually looks like (and what it misses)

Most founders I talk to assume a serious monitoring setup has to cost a serious amount of money. It doesn't. The cheapest defensible stack I see small shops actually run costs about twenty dollars a month, assembled from four free or near-free tools, and it does catch the loud failure mode — the homepage goes down, you get an email. The thing that quietly surprises people is how many failure modes that stack doesn't catch, and how much revenue leaks through the gap.

The four-piece $20/month stack

First, the UptimeRobot free tier. It pings your home page from a handful of probes every five minutes, sends an email when the page goes 5xx for two checks in a row, and sends a second email when it comes back. It is the cheapest uptime monitor that exists and it is genuinely good at the one job it does. The free tier covers fifty monitors; the paid tier is roughly seven dollars a month for the limits most small shops need. The full UptimeRobot comparison walks through where the free tier stops being enough.

Second, the Better Stack community tier. Better Stack's free community plan gives you ten monitors at a one-minute cadence from multiple regions, plus a public status page you can point customers at when something is flapping. The product is solid. The price is free for the tier most small shops will use for the first year. The Better Stack comparison walks through where the free tier stops being enough and what the upgrade actually buys you.

Third, a free SSL notifier. There are several — SSL Labs, a few open-source probes you can self-host, the no-op cert-expiry check that comes baked into your hosting control panel if you squint. The job is the same: email me thirty days before my certificate expires so I have time to renew it without taking the site down. The job is small, the tools are mostly free, and most founders either skip it entirely or set it up once and forget to look at it.

Fourth, manual plugin and dependency updates. The shop runs WordPress, or a Next.js app, or a Shopify theme. The shop remembers to log in every few weeks and update the plugins, the Node packages, the theme files. The work takes an hour. The work is the kind of work that gets pushed a quarter, then two quarters, then it doesn't happen at all.

Add it up. Seven dollars for UptimeRobot, zero for Better Stack community, zero for an SSL notifier, zero dollars but real hours for the manual updates. Twenty dollars a month, give or take. That is the stack.

What that stack catches

The stack catches the loud failure mode. The homepage returns 500 for two consecutive probes, your inbox lights up, you call your host, the host fixes it. The cert expires in twenty-eight days, your inbox lights up, you renew the cert. Both failures are loud, both are visible, both are the kind of failure the four tools were built to find. For these two failure modes, twenty dollars a month is genuinely enough.

What that stack misses

The stack misses almost everything that costs a small shop money in 2026, because most of the revenue-killing failures are not loud. They are quiet, they are partial, and they look like the site working. Here is the list I walk through with founders when they ask me whether their four-piece stack is enough.

First, the multi-site agency workflow. If you run more than three client sites — and most agencies, most in-house marketing teams, and most freelance operators I talk to do — the four-piece stack stops scaling the day you add the fourth site. Every new site needs its own UptimeRobot monitor, its own Better Stack status page, its own SSL notifier, and its own manual-update checklist. The fourth site doesn't cost twenty dollars; the fourth site costs eighty, and the tenth site costs two hundred, and the operator loses the thread. The small-business monitoring pillar walks through what the multi-site agency workflow needs to look like.

Second, the broken-link crawl. Your homepage still returns 200. Your checkout still answers. The link off your pricing page to the SaaS API you reference just shipped v2 of, and the v1 endpoint now returns a deprecation banner that breaks the embedded signup button. The visitor sees the deprecation banner, the visitor leaves, your UptimeRobot inbox stays quiet. The shape of this failure is exactly what the broken-link monitor in a layered stack catches, and it is the failure mode the four-piece DIY stack was never built to find.

Third, the checkout-watch. Your checkout page renders, the cart loads, the visitor fills in their card, and the third-party payment processor quietly returns a 500 for one in fifty transactions. The visitor sees a generic error page and assumes the site is broken. You see nothing. The four-piece stack pings the URL and sees a 200; the URL is up, the URL is healthy, the checkout is half-broken in a way the URL ping will never see. The shape of this failure is what the checkout-watch in a layered stack walks the visitor path to find.

Fourth, the form-watch. The contact form renders, the visitor fills it in, the visitor clicks submit, and the form posts to a webhook URL that quietly changed names in your CRM's last release. The visitor sees a green thank-you. You see nothing. The lead never lands in your CRM. The shape of this failure is what the form-watch in a layered stack crawls to find.

Fifth, the plugin and SQL maintenance. The fourth piece of the DIY stack — manual updates — is the piece that quietly stops happening. The WordPress plugins drift three majors behind. The Node packages drift four minors behind. The MySQL tables never get OPTIMIZE TABLE-ed, and the slow-query log fills up with joins that took forty milliseconds six months ago and now take four seconds. The site stays up. The site gets slower. The visitor blames the site. The four-piece stack never fires.

What a layered stack actually looks like

The layered stack is the four DIY pieces plus five layered monitors that close the gap. Uptime and SSL stay where they are — UptimeRobot free tier and Better Stack community are genuinely good at the loud failure mode. On top, a broken-link crawl walks the outbound links off every page a visitor would land on. A checkout-watch walks the cart path end-to-end on a one-minute cadence. A form-watchposts a synthetic form submission to every form on every page and checks that the lead lands somewhere real. A multi-site dashboard rolls every site into one inbox so the operator isn't checking five UptimeRobot tabs at 7am. And a monthly plugin / package /OPTIMIZE TABLE cadence closes the manual-update drift that quietly becomes the slow-failure story over two quarters.

The layered stack costs more than twenty dollars a month. It also closes the seven failure modes the DIY stack leaves open, which is the trade-off founders actually have to make. If the only failure you care about is the loud one, the DIY stack is enough. If the failure modes that quietly cost you revenue matter too, the layered stack is the honest answer.

When the DIY stack is the right call

The DIY stack is the right call when you run one site, when your checkout is on a first-party processor that you control, when your forms post to a webhook you can read the logs of, and when you actually remember to do the monthly plugin update. For a solo founder running a single landing-page-and-blog shop, twenty dollars a month genuinely is enough.

When it isn't

The DIY stack stops being the right call the day you add a second site, the day your checkout goes through a third-party processor you can't tail logs on, the day your CRM starts hosting the form webhook instead of you, or the day you realise you haven't logged into WordPress in three months. For those shops, the layered stack is the honest trade-off. The Sentry comparisonwalks through why a developer-grade error tracker isn't the layered stack either — Sentry is built for the engineering team, not the founder inbox.

See it on your own shop

Run the free URL auditon your home page. It pings the page, reads the cert chain, crawls a handful of outbound links, and reports what it sees in under thirty seconds. The report tells you, in plain English, which of the seven failure modes above is worth covering first — and whether the four-piece DIY stack is honestly enough for the site you actually run today.