Skip to content

For dev teams

CI passed. The deploy completed. Your pricing page just broke.

A green CI build means the code compiled — not that the live page still works for a real visitor. Sitewatch verifies production the moment your deploy lands — every page, and everything those pages ask for — and tells you in plain English what broke. No test scripts to write, and none to maintain six months from now.

  • A deploy hook fires the check the instant your pipeline finishes
  • Nothing to script — every file each page loads is discovered automatically
  • Alerts say what broke and what a user would hit, not just which request failed
No test scripts: to maintain1 site: free forever

Free scan

Scan your production site right now — free

See what your test suite never catches.

02What can go wrong

Everything upstream was green.

Your suite runs against your code, in an environment you control. Production is a different machine, a different CDN, different routing rules and a third party you do not own.

  1. 01Tests pass, the deploy completes, and the dashboard turns green.
  2. 02The page still answers, so the uptime check stays green too.
  3. 03But the served page asks for a bundle that no longer exists, so nothing on it works.
  4. 04Nothing in your pipeline ever loaded that page from production.
  5. The first check against production is a user hitting it.

03Why traditional monitoring misses this

CI verifies your code. Uptime verifies your server. Neither verifies the page.

Both signals are real, and both stop short of the thing your users touch. We don't just check that the site is online — we check that people can actually use it.

CI + uptime check

Tests pass
Server answers
Healthy

Neither one loaded a production page and its dependencies.

Sitewatch

Opens each page
Bundles load
Styling loads
Third-party scripts load
Redirects resolve
Healthy

Against production, after the deploy is live.

04How far each check goes

Why CI stops where it does.

This is not a criticism of your pipeline. Each rung costs more than the one above it, and CI is deliberately designed to stay cheap and fast — which is exactly why the bottom rungs need something that runs after the deploy.

  1. Unit tests pass

    Seconds

    Your logic is correct in isolation. Runs in seconds and never touches infrastructure.

  2. Integration tests pass

    Minutes

    Your services agree with each other, in an environment you fully control.

  3. The build and deploy succeed

    One pipeline

    The artefact was produced and shipped. Nothing has yet asked whether the result is usable.

  4. Production answers

    One request

    An uptime check confirms the server responds on the real domain. This is where ping-based monitoring stops.

  5. Most uptime tools stop here

    Every page resolves and every file it needs is served

    Many requests

    Open each page and fetch everything it references, from the real CDN, on a schedule. Dozens of requests per page — which is why cheap tools do not do it.

  6. The content is what you shipped

    A baseline per page

    Compare each page against its baseline so a stale edge, a half-published deploy or an unintended content change is visible.

The two rungs your pipeline cannot afford are the two your users start at.

05How Sitewatch catches it

One question, after every deploy.

Not "did the pipeline pass?" — but the one production answers for you otherwise: is the deployed page usable right now?

  • Every monitored page resolves
  • Bundles are served
  • Styling is served
  • Fonts and images load
  • Third-party scripts load
  • Redirects resolve correctly
  • Content matches its baseline
  • Certificate is valid

06Typical failures we catch

The failure surface CI does not cover.

The CDN is not serving a file the page asks for

A path or hash changed and the served page still references the old one.

The page loads and nothing on it works

A file comes back as the wrong kind of file

A host or cache layer serves an error page where a script should be.

The browser refuses it silently and the feature disappears

A third-party script stopped being served

Auth, analytics or chat, from infrastructure you do not control.

Their outage becomes your broken login

A routing rule creates a redirect loop

Platform-level rules collide in a way no test environment reproduces.

One route is unreachable while the homepage looks perfect

One edge is serving a stale build

The deploy published, but not identically everywhere.

A region gets a mismatched app and your dashboard averages it away

Content changed without anyone shipping it

A CMS edit, a cache purge, or something less pleasant.

The page drifts from what you believe is deployed

07Fits your pipeline

Running in under five minutes

  1. 01

    Add the deploy hook to your pipeline

    Copy the Sitewatch webhook URL and add it as a deploy notification in Vercel, Netlify, GitHub Actions, CircleCI, or anything that can make an outbound request. Takes under a minute.

  2. 02

    Your deploy fires it

    The moment the deploy completes, the hook triggers a check against production. No polling, no waiting for the next scheduled interval.

  3. 03

    Every page and everything it loads is verified

    Sitewatch walks the pages you monitor, fetches every file they reference from the real CDN, and compares each page against its baseline.

  4. 04

    A verdict lands in Slack

    Within minutes: what passed, what failed, what changed since the last check, and a plain-English description of anything that broke.

08What happens the moment we find it

Detection is not the outcome. Rolling back is.

The alert is specific enough that the decision to revert takes seconds instead of a bisect.

We confirm it is real first

Every issue is re-checked before anyone is paged, so a single slow response mid-deploy does not wake the on-call engineer.

Diagnosed in plain English

What broke, which file, and what a user would experience — useful to whoever is on call, not just to whoever wrote it.

Ranked by what it costs

A broken checkout and a missing blog footer are not the same incident. Severity reflects the page, so a broken image on the blog never crowds out a broken checkout.

You can verify the rollback

Revert, run a check on demand, and close the incident on evidence rather than on the pipeline going green a second time.

10vs. what you already run

Your CI went green. That does not mean production works.

Files missing from the real CDN

CI/CD alone:
Not checked — tests never touch production
CI/CD + Sitewatch:
Every file fetched from production after the deploy

A file served as the wrong kind of file

CI/CD alone:
Silent — tests do not inspect what the host returns
CI/CD + Sitewatch:
Verified on every file, on every check

Third-party script outages

CI/CD alone:
Not covered — usually stubbed out in tests
CI/CD + Sitewatch:
Checked in production alongside your own files

Test scripts to write and maintain

CI/CD alone:
Yes — and they rot the moment the UI changes
CI/CD + Sitewatch:
None. Everything the page loads is discovered

Missing styling

CI/CD alone:
Only if someone wrote a test for it
CI/CD + Sitewatch:
Automatic, on every check

Platform-level redirect loops

CI/CD alone:
Rarely — hosting rules are not in your test env
CI/CD + Sitewatch:
Redirect chains followed and flagged

What the alert reads like

CI/CD alone:
A stack trace, needing interpretation
CI/CD + Sitewatch:
What broke, where, and what a user would hit

Anything checks production after deploy

CI/CD alone:
No — CI passes before the deploy and stops
CI/CD + Sitewatch:
Yes — the deploy hook triggers it

Webhooks

Deploy hooks for any CI/CD platform

Minutes

From deploy to verdict

0 scripts

No browser tests to write or maintain

See what your last deploy actually did

Free plan monitors 1 site. Pro covers 100 sites at $19/month.

11Questions

Dev team questions, answered