Keyword Monitoring - Catch Broken Pages That Return 200 OK

Keyword monitoring is an uptime check that downloads a page and verifies that a specific piece of text is present in the response. A plain HTTP monitor only looks at the status code, so it reports "up" whenever the server answers with 200 OK, even if the page shows a database error or a blank template. A keyword monitor fails in those cases, because the text you expect is missing. Pick a word or phrase that only appears when the page is working end to end, such as a button label rendered from your database.

Failures that return 200 OK

Plenty of real outages do not produce an HTTP error code:

Content management systems, frameworks and hosting control panels often show friendly error pages with a 200 status, which is exactly what a status-only monitor cannot detect.

Choosing a good keyword

The keyword is the most important setting. A good one is:

1. Specific to the working page. It should not appear on error, maintenance or login pages.
2. Generated by your application, ideally from the database. A product name or a list heading that comes from a query proves the database is answering.
3. Stable. Avoid text that marketing changes weekly, prices that change daily, dates and session-specific values.
4. Present in the HTML source, not only after JavaScript runs.
5. Short but not generic. "Add to cart" on a product page is good. "Home" is not, because it appears on almost every page, including error pages.

Good and bad examples

See what the monitor sees

Before you save a keyword, confirm it is in the raw response:

curl -sL https://example.com/product/blue-mug | grep -c "Add to cart"

A result of 1 or higher means the text is in the HTML. A result of 0 means either the text is rendered by JavaScript or it is spelled differently in the source.

To view the response headers and status as well:

curl -sIL https://example.com/product/blue-mug

Some servers return different HTML to different user agents or block requests that look automated. If the page looks different to curl than to your browser, check firewall and bot-protection rules, and allow your monitoring service if necessary.

Build a health endpoint for APIs and apps

For an application you control, the most reliable keyword target is a small endpoint that tests real dependencies and reports the result in plain text or JSON:

GET /health

{"status":"ok","database":"ok","cache":"ok"}

Guidelines:

• Run a cheap query, such as SELECT 1, rather than a heavy report.
• Return a non-200 status when a dependency fails, so both the status check and the keyword check catch it.
• Keep it fast, under a second, so the check does not time out.
• Do not include secrets, version strings with known vulnerabilities or internal hostnames in the output.

Then monitor /health with the keyword "database":"ok".

Inverse keywords: alerting when text appears

Some monitoring tools also support the opposite rule: fail if a word is present. That is useful for catching:

• Fatal error or Stack trace leaking onto a page
• maintenance on a page that should be live
• Known spam words injected into a hacked site

If your tool does not support inverse matching, you can usually achieve the same result with a positive keyword that disappears when the page breaks.

Keyword monitoring checklist

• Monitor the pages that earn money or generate leads, not just the homepage
• One keyword per monitor, unique to the working state of that page
• Keyword confirmed with curl in the raw HTML
• Avoid prices, dates, personalized content and frequently edited marketing copy
• Use a health endpoint for apps and APIs
• Re-check keywords after redesigns, CMS migrations and theme updates, because they often change labels
• Set a timeout of 10 to 30 seconds so a very slow page also counts as a failure

Common causes of false alarms

• A/B tests that change button text for some requests
• Geo or language redirects that serve a translated page to the monitoring location
• Cookie consent walls that replace content for first-time visitors
• Rate limiting or bot protection that returns a challenge page
• Cache purges that briefly serve a fallback page

If a keyword monitor flaps, fetch the page with curl from a different network and compare the output before changing the keyword.

PingRobot offers keyword monitors next to plain HTTP(S) checks, and confirms failures from a second region before alerting. For the basics of monitoring, read what is uptime monitoring, and if a check fails, work through the website down checklist.

Start monitoring free — 10 monitors, no credit card.