Monitor Competitor Prices & Get AI Alerts with n8n

The Problem with Manual Price Checking

Checking three competitor URLs every morning sounds manageable. Then you add a fourth product line, a second market, and a weekend shift. Within a month you're either paying for an expensive SaaS repricing tool or silently losing margin because you missed a price drop on a Tuesday.

An n8n workflow fixes this for pennies. You define the URLs once, tell n8n when to check them, and let an AI node do the interpretation — not just "price changed" but "Competitor B dropped their Pro plan 18% the day before Black Friday; consider matching."

This guide walks through the exact nodes, a real config, and how to run it on your own always-on n8n instance that already has AI credits included.

---

What You Will Build

A scheduled workflow that:

1. Pulls competitor pricing pages on a fixed schedule (every hour, or daily at 08:00).
2. Extracts the relevant price from each page with an HTTP Request + HTML Extract node.
3. Stores the latest known price in a Google Sheet so you have a comparison baseline.
4. Passes the old price, new price, and product name to an AI/LLM node.
5. The AI node writes a plain-English summary of what changed and whether it looks significant.
6. Sends the summary to Slack, email, or a Telegram chat — only when something actually changed.

---

The Node-by-Node Build

1. Schedule Trigger

Add a Schedule Trigger node. Set it to Every Hour (or Every Day at 08:00 UTC). This is the entry point — n8n wakes up on its own, no cron server needed.

Trigger mode: Interval
Every: 1
Unit: Hours

2. Code Node — Define Targets

Use a Code node (JavaScript) to return an array of competitor URLs and product identifiers. Keeping targets here (not in a spreadsheet) makes the workflow readable.

return [
  { competitor: "AcmePro", url: "https://acmepro.io/pricing", selector: ".price-hero .amount" },
  { competitor: "BoltSaaS", url: "https://boltsaas.com/plans", selector: "#monthly-price" },
  { competitor: "ZenTool",  url: "https://zentool.co/pricing", selector: "[data-price]" },
];

3. Split In Batches

Drop a Split In Batches node after the Code node with Batch Size: 1. This lets the rest of the workflow run once per competitor — clean and easy to debug.

4. HTTP Request — Fetch the Page

Add an HTTP Request node:

• Method: GET
• URL: {{ $json.url }}
• Response Format: Text

No auth needed for public pricing pages. If a site blocks the default User-Agent, add a header: User-Agent: Mozilla/5.0 (compatible; PriceBot/1.0).

5. HTML Extract — Pull the Price

Add an HTML Extract node:

• Source Data: {{ $json.data }} (the raw HTML from step 4)
• Extraction Values:
  • Key: price
  • CSS Selector: {{ $('Split In Batches').item.json.selector }}
  • Return Value: Text

This gives you { price: "$49" } (or whatever text is in the element).

Tip: Always clean the extracted string with a Set node: {{ $json.price.replace(/[^0-9.]/g, '') }}. Stripping currency symbols lets you do numeric comparisons later.

6. Google Sheets — Read Last Known Price

Add a Google Sheets node in Read mode. Keep one sheet with columns: competitor, lastPrice, lastChecked.

• Operation: Read Rows
• Filter: Column competitor equals {{ $('Split In Batches').item.json.competitor }}

This returns the stored baseline. If no row exists yet (first run), the output will be empty — handle that with an IF node downstream (treat "no row" as "price is new").

7. IF Node — Did the Price Change?

Add an IF node with one condition:

{{ $('HTML Extract').item.json.priceClean }} != {{ $('Google Sheets').item.json.lastPrice }}

• True branch → send to AI node + alert.
• False branch → Stop (no noise, no alert).

8. AI / LLM Node — Interpret the Change

This is where the magic happens. Add an AI node (the built-in LLM node in n8n) and set the System Prompt:

You are a pricing analyst. You receive a competitor name, their old price, and their new price.
Write a 2–3 sentence plain-English summary: what changed, the percentage difference, and one
actionable observation (e.g., "this may signal a promotional push" or "within normal A/B range").
Be direct. No fluff.

User message:

Competitor: {{ $('Split In Batches').item.json.competitor }}
Old price: ${{ $('Google Sheets').item.json.lastPrice }}
New price: ${{ $('HTML Extract').item.json.priceClean }}

On AgentRoost, this node works immediately — no OpenAI key to paste, no billing page to configure. The AI credits are bundled into your plan.

9. Google Sheets — Write Updated Price

Add another Google Sheets node in Update mode to overwrite the row for this competitor with the new price and a timestamp. Now next hour's run will use today's price as the baseline.

10. Slack (or Email / Telegram)

Add a Slack node or Send Email node:

• Channel / To: your team channel or your inbox
• Message: {{ $('AI').item.json.text }}

If you prefer Telegram, use the Telegram node with your bot token and chat ID. The message body is the AI node's output — already human-readable, no template to write.

---

Handling Edge Cases

Site changed its HTML structure — The selector stops matching and price comes back empty. Add a second IF branch: if priceClean is empty → send "Selector broken for [competitor]" alert so you notice before silently losing coverage.

Price temporarily missing (429 / maintenance page) — Wrap the HTTP Request in an Error Trigger or use the node's built-in retry settings (Max Tries: 3, Wait Between Tries: 60s). Only write to Sheets if the request succeeded.

Multiple price tiers — Run the Code node with an array that includes a tier field ({ competitor: "AcmePro", tier: "Pro", selector: ".pro .amount", url: "..." }). The IF node compares per tier, and the AI node gets the tier in its prompt.

---

How to Run This on AgentRoost

You need an always-on instance — not a laptop you close at night, not a cron job that lapses when your VPS restarts. This is where self-hosting collapses for most people: the maintenance overhead.

On AgentRoost you get your own n8n instance, running 24/7, at https://<your-id>.agentroost.app — your login, your workflows, your data. The instance never sleeps. And the AI node in step 8 already has credits loaded; you pick from 350+ LLM models and switch anytime.

Setup takes about 2 minutes:

1. Sign up at agentroost.app/en/agents/n8n
2. Pick the n8n framework, name your instance
3. Your private n8n editor opens at https://<your-id>.agentroost.app
4. Import or build the workflow above — the AI node will have credits ready

No Docker. No SSL certificate. No API key to paste for the LLM. All-in from $19.99/month with a 14-day money-back guarantee.

Compare plans and get started →

---

Tips for a Production-Grade Monitor

• Run every hour during business hours, daily at night. Use the Schedule Trigger's advanced cron field: 0 8-18 * * 1-5 (weekdays, 08:00–18:00 UTC hourly).
• Archive history, not just last price. Add an extra Google Sheets Append node to a second "history" sheet so you can chart price trends over weeks.
• Throttle the batch. If you have 20+ competitors, add a Wait node (2–5 seconds between requests) to avoid looking like a bot storm.
• Test selectors with a Webhook trigger first. Create a manual Webhook node that returns a hardcoded HTML snippet to iterate on your CSS selectors without hitting live sites repeatedly.