OnWebChange CSS selector tracking, and when a sentence works better

OnWebChange CSS selector tracking is the reason many developers picked the service in the first place: point a selector at one element, and only that element gets compared. This explainer covers how the selection, rules and number conversions fit together, where the approach breaks, and when writing the condition as a sentence in a tool like AyeWatch is the better trade.

How the selection works

Every OnWebChange plan, including the free one, includes what the vendor calls advanced CSS-based content selection. You create a tracker for a URL, then narrow it to the part of the page you care about with a CSS selector. A single tracker can watch several elements on the same page, so a product tracker might select the price, the stock label and the shipping note at once.

OnWebChange renders JavaScript pages, so selectors work on content that loads after the first HTML response. It can also send custom headers to reach pages behind a login, and lets you choose tracking locations in different regions, which helps when a site serves different content by country.

Rules, filters and number conversions

Selection decides what is compared. Rules decide whether a difference becomes an alert. OnWebChange includes rules, filters and number conversions on every plan:

• Notification rules hold an alert back until specific conditions or keywords appear.
• Number conversion treats text like "€1,249.00" as a value, which suits prices and exchange rates.
• Filters narrow what counts as a change.

On Standard and Turbo plans, numeric trackers also get history drawn as line graphs, which is handy for watching a price over months. OnWebChange also lists AI summarization and AI translation, so an update can be condensed or read in another language.

Where selectors break

The weakness is the same one every scraper has. A selector encodes the page's structure, and structure changes:

• A redesign renames classes or wraps elements in new containers. The selector matches nothing, or matches the wrong block.
• A/B tests serve different markup to different visitors.
• Content moves to a new component, and the old one stays on the page but empty.

In each case the tracker either goes quiet or fires on something you did not mean. Fixing it means opening the page, finding the new path, and saving again. For five trackers that is fine. For fifty it becomes a chore.

Keyword rules have a quieter version of the same problem. They test whether a string is present, not what the edit means. "Sold out" appearing in a reviews block counts the same as "sold out" next to the buy button.

The sentence approach

AyeWatch drops the selector entirely. You paste one or more URLs into a webpage monitor and write the condition in plain English:

"Alert me when the minimum supported Node version changes on this install page. Ignore edits to the troubleshooting section."

AyeWatch judges each change against that rule and sends a short summary when it matches. Nothing points at a DOM path, so a redesign does not break the monitor.

For a developer, the paid plans add the parts that make it scriptable: a REST API to manage topics, HMAC-signed webhooks with the summary in each event, and an MCP server so a coding agent can set up a monitor for you. Alerts go to the iOS and Android apps and the in-app feed on every plan; Slack, Discord and email come with paid plans.

What you give up

Precision. A CSS selector on one number will always be more exact than a sentence, and a rule on a converted number is hard to beat for a single price. AyeWatch also covers public pages only, so the custom-header trick for login pages has no equivalent, and it does not list PDF tracking.

Picking per tracker

A practical split:

AyeWatch has a Free Preview and paid plans from $9 a month, and it sends alerts to the web, iOS and Android apps, with Slack, Discord and email on paid plans. The OnWebChange comparison has the full feature table.