Scrape Web Pages: Claude + Make

A used-bookstore owner keeps an eye on three rare-book listing pages, hoping to catch a first edition before someone else does. Checking them by hand means opening tabs every morning, squinting at prices, and usually forgetting by Thursday. The information is public and structured; the chore is the watching.

This build does the watching. On a schedule you set, Make fetches the page, Claude reads it and pulls out the handful of facts you care about, and each check lands as a row in a Google Sheet. Over a week you have a dated log you can sort and chart, built while you slept.

We wire it in Make, and it is the first build in our data and research set. A note up front: only scrape pages you are allowed to, respect a site’s terms and its robots file, and keep your schedule gentle. This is for public information, gathered politely.

What does this workflow actually do?

In one line: a web page becomes a row in a sheet. On the schedule you choose, Make fetches the page over HTTP, hands the text to Claude with a list of the facts you want, Claude returns them cleanly, and Make appends a timestamped row. No tabs, no copy-paste, just a growing log.

A few real cases, none of them the usual ones:

• The rare-book hunter above, catching a price drop the morning it happens.
• A community theater volunteer watching a grants page for new funding calls.
• A beekeeping-supply buyer tracking a wholesaler’s stock page for restocks.
• A high-school robotics team watching a parts vendor’s clearance page before a build season.

Reading one page is easy. Reading it every day, forever, and remembering what changed is the chore. That patient watching is exactly what an automation does without complaint.

Why use Make if Claude can already read and write?

Fair question. Claude is good at reading a messy page and pulling out clean facts. It is not built to fetch a page on a schedule, run unattended, and append rows to the right sheet. That is plumbing, and Make does plumbing.

Make handles the no-judgment steps: fetching on schedule, passing the text along, writing the row, running on its own. Claude handles the one step that needs reading, the extraction. Make is the timer and the courier; Claude is the one that actually reads the page. I tried asking the AI to do the fetching and scheduling too, and it was slower and broke whenever a page shifted. Splitting it keeps the log reliable.

What do you need before you start?

• A Make account. The free plan covers 1,000 operations a month, plenty for a daily check on a few pages.
• A Claude API key from the Claude console, not the chat app. New to Claude? Start with how to use Claude.
• A Google Sheet with the columns you want to capture, plus a Date column.
• A public page that allows automated access. The HTTP step needs no account; it just fetches the page.

Two Make words, once. A scenario is one automation, the whole fetch-to-sheet chain. A module is one box in it. Three boxes, one scenario.

How does the workflow work, step by step?

Three modules, left to right:

Module	App	What it does
1. Trigger	HTTP	Fetches the page on the schedule you set
2. Brain	Claude	Reads the page and pulls out the facts you defined
3. Output	Google Sheets	Appends the facts as a timestamped row

Step 1: Fetch the page on a schedule

Create a scenario and add HTTP, Make a request as the first module. Set the method to GET and paste the page URL. Then set the scenario’s schedule (the clock icon) to something gentle like once a day. On each run, this module downloads the page so Claude has the content, not just the address. Keep the schedule reasonable so you are a good guest on someone else’s server.

Step 2: Let Claude pull the facts

Add a Claude module (Anthropic), choose Create a Prompt, paste your API key, and pick Claude Sonnet. Tell it exactly what to extract:

Extract these facts from the page below. Return them on one line,
separated by a pipe, and nothing else. If a fact is missing, write
"not found" for it, never guess.
Page: {{data from Step 1}}

Title | Price | Availability

Swap in whatever fields you track. The “not found” rule keeps Claude from inventing a price, and the pipe format makes the next step line up cleanly.

Step 3: Log it to your sheet

Add Google Sheets, Add a Row, connect your account, pick the sheet, and map Claude’s pipe-separated facts into your columns, plus the current time into a Date column. Now every run leaves a dated row, and a week of rows is a chart you can read at a glance.

How do you run it and check the result?

Click Run once. Make fetches the page, Claude pulls the facts, and a row appears in your sheet. Check it. When I tested this against a simple listing page, the title and price parsed correctly on the first try, and the fix that mattered was telling Claude to write “not found” for a missing field instead of guessing one. Tune the Step 2 fields, then turn the scenario on with your daily schedule.

After that, a page you used to babysit logs itself. You stop checking and start reading a history, which for prices, stock, or postings is where the useful patterns live.

What does this cost to run?

Piece	Free tier	If you outgrow it
Make	1,000 operations/month free	Core plan from about $9/month
Claude API	Pay per use	A short extraction on Sonnet costs a fraction of a cent
Google Sheets	Free	Free
HTTP	Free	Free

A daily check on a few pages is a few dozen operations a month, deep inside the free Make plan. For the full tier breakdown, see our Make beginner’s guide.

What can go wrong, and how do you avoid it?

• The fetch returns junk or nothing. Some pages block simple requests or load through scripts; those need a heavier scraper. Most static pages are fine.
• Claude invents a missing value. The “not found” rule in the prompt stops that. Keep it.
• You scrape too aggressively. A gentle schedule respects the site and your operation count. Daily beats every minute.
• You log a page you should not. Check the site’s terms and robots file first; scrape only public, permitted pages.

Want more of this pattern? Our Make AI scenarios roundup has the next ideas.

How do you build this in Zapier or n8n instead?

The three jobs do not change, only the names do.

Job	Make	Zapier	n8n
Fetch the page	Make a request (HTTP)	Webhooks / Code by Zapier	HTTP Request node
Pull the facts	Claude module	Claude (Anthropic) action	Anthropic node
Log the row	Add a Row	Create Spreadsheet Row	Google Sheets node

Make and Zapier are the gentlest for a first build. n8n suits self-hosting and high volume. We compared all three in Zapier vs Make vs n8n.

Common questions

Is web scraping allowed?

Scraping public information is common, but you must respect each site’s terms of service and its robots file, and keep your requests gentle. When in doubt, do not scrape it.

Why not just ask Claude to read the URL?

The chat app cannot reliably fetch live pages, and the API model needs the text given to it. The HTTP step downloads the actual page so Claude has something real to read.

Can it handle pages that need a login?

Not with a plain fetch. Login-only or heavily scripted pages need a dedicated scraping tool. This build is for public, static pages.

How often should it run?

As rarely as your need allows. Daily is plenty for most prices and postings, and it keeps you a polite guest on the site you are reading.

Is the Claude chat app the same as the API?

Same models, different door. The app is for you; Make talks to the API, so you need a key from the console.

Sources and docs

• HTTP module (Make Help)
• Claude API overview (Anthropic)
• Scheduling a scenario (Make Help)
• Robots Exclusion Protocol (RFC 9309)
• Web scraping — Grokipedia

Last reviewed: May 2026. Make, Claude, and Google update their interfaces; check the official pages for exact button names.

You may also like

• AI Automation for Beginners (the hub)
• Summarize any web page into a newsletter blurb
• Draft WordPress posts from a spreadsheet with Claude + Make
• Make.com Complete Beginner’s Guide
• How to use Claude AI
• Zapier vs Make vs n8n: which to start with
• The AI Tools Directory