How to Monitor a Website for Keyword Appearances (Regex Monitoring)

Most monitoring tools answer the wrong question. They tell you "the page changed" when what you actually need is "did this specific string appear, disappear, or shift in a way that matters?" That distinction is the difference between a useful alert and an inbox full of noise.

This guide covers how to build precise, regex-powered keyword monitors using Verid's extraction and predicate system. Every config shown here is real and runnable.

Why Regex Belongs in Your Monitoring Stack

CSS selectors break when a class name changes. XPath breaks when an element moves. Regex operates on raw page source, which means it works even when the DOM has no clean structure to target.

That makes it the right tool for a handful of specific situations:

• A version string is injected into a <script> tag as a JavaScript variable
• A price or date appears inline in a paragraph with no wrapping element
• You want to count how many times a term appears on a page (sitemap URLs, external links, keyword density)
• The markup is inconsistent across the pages you need to monitor

Regex will not replace CSS or XPath for well-structured pages, but it covers the gaps those methods leave behind.

How Verid's Regex Extraction Works

Verid runs regex patterns against the full raw source of the fetched page, including all HTML tags, script contents, and embedded JSON. There are two modes depending on whether you include a capture group:

The /…/ delimiters mark a regex. Without them, Verid counts exact string matches.

JSON escaping rule: Regex backslashes must be doubled inside JSON strings. Write \\d+ not \d+. Write \\. not \.. Test your pattern at regex101.com against a page source fragment before saving.

Setting Up a Regex Monitor via the API

Every Verid monitor is created with a single POST /v1/monitors call. Here is the full shape for a regex-based keyword monitor:

POST https://api.verid.dev/v1/monitors
Authorization: Bearer vrd_your_api_key
Content-Type: application/json

{
  "name": "Terms page - last updated date",
  "url": "https://example.com/terms",
  "schedule_interval_seconds": 86400,
  "extract_config": {
    "method": "regex",
    "fields": {
      "last_updated": "/Last updated: ([A-Za-z]+ \\d{1,2}, \\d{4})/"
    }
  },
  "diff_predicate": {
    "type": "field_changes",
    "field": "last_updated"
  },
  "deliveries": [
    { "type": "webhook", "url": "https://your-app.com/hooks/terms-change" }
  ]
}

Verid runs the loop: fetch the URL, apply the regex, compare against the last stored value, evaluate the predicate, and if it fires, deliver a signed webhook with the before/after diff. You write none of that infrastructure yourself.

Regex Patterns for Common Keyword Monitoring Scenarios

Version number in a script tag

A common pattern: an app embeds its version into the page as a JS variable.

Target HTML:

<script>
  window.__CONFIG__ = { version: "3.14.2", env: "production" };
</script>

Extract config:

{
  "method": "regex",
  "fields": {
    "version": "/version: \"(\\d+\\.\\d+\\.\\d+)\"/"
  }
}

Predicate - fire on any version bump:

{ "type": "field_changes", "field": "version" }

Returned value: "3.14.2"

Stock status keyword

Watch for an exact phrase transitioning in or out. Useful for product restock monitoring.

Extract config:

{
  "method": "regex",
  "fields": {
    "availability": "/(In Stock|Out of Stock|Backordered)/"
  }
}

Predicate - alert only when it reads "In Stock":

{
  "type": "field_matches_regex",
  "field": "availability",
  "pattern": "^In Stock$"
}

This fires only on a match, not on every page crawl. See the full change detection predicate reference for all nine predicate types.

Price extracted from inline paragraph text

No price element, just a number buried in a sentence.

Target HTML:

<p>The annual plan is currently priced at <strong>$149.00</strong> per seat.</p>

Extract config:

{
  "method": "regex",
  "fields": {
    "annual_price": "/\\$([\\d,]+\\.\\d{2})/"
  }
}

Returned value: "149.00" (the capture group excludes the $ sign)

Predicate - alert when the price drops by 5% or more:

{
  "type": "field_decreases_by_percent",
  "field": "annual_price",
  "threshold": 5
}

Keyword occurrence count

Count how many times a keyword appears across a page, no capture group needed.

Extract config:

{
  "method": "regex",
  "fields": {
    "keyword_count": "data-privacy"
  }
}

Returned value: 14 (integer count of matches)

Predicate - fire if the count drops to zero (keyword removed):

{
  "type": "field_equals",
  "field": "keyword_count",
  "value": "0"
}

Regex Pattern Reference

Combining Regex Extraction with Composite Predicates

Where regex monitoring gets genuinely powerful is in combination with composite AND/OR predicates. Here is a monitor that fires only when a competitor's pricing page shows a price drop AND the item is in stock:

{
  "name": "Competitor: price drop on in-stock item",
  "url": "https://competitor.com/product/widget-pro",
  "schedule_interval_seconds": 900,
  "extract_config": {
    "method": "regex",
    "fields": {
      "price": "/\\$([\\d,]+\\.\\d{2})/",
      "availability": "/(In Stock|Out of Stock)/"
    }
  },
  "diff_predicate": {
    "type": "composite",
    "operator": "AND",
    "conditions": [
      {
        "type": "field_decreases_by_percent",
        "field": "price",
        "threshold": 5
      },
      {
        "type": "field_equals",
        "field": "availability",
        "value": "In Stock"
      }
    ]
  },
  "deliveries": [
    { "type": "webhook", "url": "https://your-app.com/hooks/repricer" },
    { "type": "slack" }
  ]
}

One config, zero polling loops, no false positives on out-of-stock items.

How This Compares to Other Monitoring Approaches

The core gap with screenshot tools is they alert on any pixel change. A cookie banner update, a rotating ad, a timestamp in the footer: all of those fire. Predicate-driven monitoring only fires when the condition you defined is true.

Common Issues and Fixes

FAQs

Does regex extraction work on JavaScript-rendered pages?

Yes. Verid's fetcher auto-escalates: it tries a static fetch first, then falls back to a headless browser, then to a residential proxy if the site blocks bots. The regex runs on whatever source the fetcher returns. If you need JS-rendered content, it is handled without any extra config.

Can I monitor for a keyword appearing OR disappearing?

Yes. Extract the keyword as a plain string count (which returns an integer). Then use field_equals with a value of "0" to trigger when the keyword has been removed, or use field_increases_by_absolute with a threshold of 1 to trigger the first time it appears.

What happens when a site restructures its HTML and breaks my selector?

If a CSS or XPath selector breaks, you can switch the extraction method to regex or to AI extraction with a config update, no redeployment required. The LLM extractor lets you describe the field in plain English as a fallback.

How do I test a regex pattern before creating a monitor?

Paste a fragment of the page source into regex101.com and verify that your capture group returns exactly what you expect. Then double all backslashes when you move the pattern into the JSON config. The Verid regex extraction guide walks through five complete examples with inputs and expected outputs.