How to monitor a website for changes and get an alert
This guide first appeared on usewrit.app. I build Writ, the tool in the last part; the first three methods need nothing from us. A supplier updates a price list. A competitor quietly changes its pricing page. A school publishes its admission dates, a status page flips to "degraded", a job board post

s the role you were waiting for. In every case the information was public, and in every case you found out late, because nobody tells you when a web page changes. This guide covers the ways to monitor a website for changes, what each one costs you in time, and how to set up an alert that fires only when the part you care about moves. Almost every page changes on every load: a date in the footer, a rotating banner, a visitor counter, an ad.
If you compare the whole page, you get an alert every time and you stop reading them within a week. Good change monitoring starts by narrowing the target to the part that matters: An element (a CSS selector): the price, the status label, the table of dates, one paragraph of the terms. A visual zone: a region of the rendered page, compared as pixels, which catches layout and image changes that text misses. A screenshot region: the same idea for pages where the content is drawn rather than written, such as a chart or a map.
Then decide what counts as a change worth an alert: any difference, a number crossing a threshold, a word appearing ("in stock", "open", "approved"), or a word disappearing. A bookmark and a calendar reminder. It costs nothing and works for one page you visit weekly anyway. It fails the moment the timing matters, because you see the change when you look, not when it happens.
Extensions reload a tab on a timer and highlight differences. They are quick to set up, but they only run while your browser is open, on that one computer, and they compare what the page shows you, banners and all. Fetch the page, extract the element, compare it with the last value, send a message. A developer can write this in twenty lines: import json, pathlib, requests from bs4 import BeautifulSoup URL, SELECTOR = "https://example.com/pricing", ".plan-pro .price" state = pathlib.Path("last.json") html = requests.get(URL, timeout=20).text now = BeautifulSoup(html, "html.parser").select_one(SELECTOR).get_text(strip=True) before = json.loads(state.read_text())["value"] if state.exists() else None if before is not None and now != before: requests.post(WEBHOOK_URL, json={"text": f"{URL} changed: {before} -> {now}"}) state.write_text(json.dumps({"value": now})) The script is the easy part.
The work is everything around it: a server that stays on, pages that only render their content after JavaScript runs (the fetch above returns an empty shell), a selector that breaks the day the site is redesigned and fails silently, pages behind a sign-in, and a log of what changed when. Multiply by the number of pages you watch. A service keeps the baseline, runs the checks on its own servers, renders JavaScript when needed and sends the alert. This is what Writ Monitors do, with two differences that matter in practice: a change can run an action, not only send a message, and a selector that breaks is repaired instead of going quiet.
Paste the URL of the page you want to watch. Pick the target: point at the element, or draw the visual zone or screenshot region. Set the condition: any change, a value above or below a threshold, or text that appears or disappears. Choose the cadence: as often as your plan allows, or as rarely as once a day.
Choose the response: an alert (email, SMS, WhatsApp, Signal, Pushover), a webhook to your own tools, or a workflow that runs on the change. Writ stores a baseline, checks the page on your schedule and records each change with the value before and after, so you also get a history of the page, not only a notification. The free page change monitor tool shows what a monitor would track on any page before you sign up. On the Free plan a page can be checked every 5 minutes without a browser, and every 15 minutes when it needs one; higher plans go down to every 10 seconds.
In Writ Cloud a check without a browser costs 0.5 credits and a browser check 5 credits; a check that could not complete costs nothing. Checks that run on your own computer with Writ Desktop cost 0 credits. New accounts start with 1,000 free credits, no card. Faster is not always better.
A price or a status page rarely needs more than one check every few minutes; a terms page or a docs page is fine once a day. Match the cadence to how fast you would act on the alert, and respect the site's rate limits. Many modern pages send an empty shell and build the content in the browser. A plain fetch sees nothing, so the check needs a real browser to render the page first: choose a browser check when the content is not in the page response.
For a page inside your own account (a supplier portal, an order status, a dashboard), attach a persona: Writ uses your signed-in session to reach the page, with your credentials resolved privately when the check runs. This is where home-made monitors die. The selector no longer matches, the script reads nothing, and the alerts stop without anyone noticing. Writ re-anchors the target as the page shifts, and when a check still fails, auto-repair rebuilds the broken step and tests it on the live page.
Repairs are free. An alert still needs a person to do the next thing. In Writ, a change is a trigger: it can send the new value to a webhook, add a row to your data, or run a workflow you recorded, such as filling in a registration form the moment a closed page opens. Conditions, a cooldown and an optional approval step keep the action inside limits you set.
Two common setups: register when a page opens and competitor monitoring. If you use Claude, ChatGPT, Cursor or another MCP client, connect Writ's MCP server and ask in plain words: "tell me when the Pro price on this page changes". The agent creates the monitor and wires the alert; the checks keep running after the chat is closed. Competitor pricing, plans and feature pages.
Stock status on a product page. Status pages, changelogs and API docs of the tools you depend on. Admissions, grants, tenders and public notices with a deadline. Job boards and careers pages for one role or one company.
Terms of service and privacy policies of your vendors. Watch pages you are allowed to access, and follow each site's terms.
Key Takeaways
- β’In every case the information was public, and in every case you found out late, because nobody tells you when a web page changes.
- β’This story was reported by Dev.to, covering developments in the dev space.
- β’AI advancements continue to reshape industries β read the full article on Dev.to for complete coverage.
π Continue reading the full article:
Read Full Article on Dev.to βShare this article



