Automate Keyword Service Detection with n8n & AI Agents

1. Opening Problem Statement

Meet Anna, an SEO specialist managing hundreds of keywords in a Google Sheet for her company’s digital marketing campaigns. Every day, she wastes hours manually filtering keywords to identify which ones correspond to known IT software, services, or applications. This painstaking task often results in missed opportunities and human errors, reducing the quality of her keyword research and overall ROI.

What if Anna could automatically detect whether each keyword in her large dataset references a known IT service or software tool? This would save her countless hours, minimize errors, and optimize her campaigns’ targeting strategy. Let’s explore how this exact challenge is solved with this powerful n8n workflow that integrates AI agents and automation.

2. What This Automation Does

This workflow efficiently processes large keyword lists from Google Sheets and intelligently classifies each keyword based on whether it relates to an IT software or service. Here’s exactly what happens when you run this workflow:

• 🔹 Keywords are fetched directly from a specified Google Sheet.
• 🔹 Keywords are split into manageable batches for processing to avoid API overload.
• 🔹 Each batch waits briefly to prevent hitting API rate limits.
• 🔹 Each keyword is sent to a custom AI agent powered by OpenAI GPT-4 that examines the keyword and decides if it matches the name of any known IT service or software.
• 🔹 The AI agent returns a structured “yes” or “no” response for each keyword.
• 🔹 The workflow updates the original Google Sheet with the AI’s classification results for each keyword, adding a new “Service?” column with the answer.

This automation not only saves Anna roughly 3-4 hours per batch run but also boosts accuracy and scalability of keyword analysis by leveraging cutting-edge AI integration.

3. Prerequisites ⚙️

• 🔑 An active n8n account or self-hosted instance (self-hosting options available at Hostinger).
• 📊 Google Sheets account with access to the specific sheet containing keywords.
• 🔐 OpenAI API account and credentials for GPT-4o-mini model access.
• ⚙️ Basic familiarity with n8n workflow editor for setup and tweaks.

4. Step-by-Step Guide

Step 1: Trigger Workflow Manually

Navigate to the n8n editor and locate the When clicking ‘Test workflow’ node (Manual Trigger). This triggers the workflow manually, ideal for testing or on-demand runs.

Click the Execute Workflow button or press “Test Workflow”. You will see the execution start from this node, confirming the trigger.

Common mistake: Forgetting to trigger manually during testing leads to no data processing.

Step 2: Fetch Keywords from Google Sheets

Next, the workflow fetches keyword data from a specific Google Sheet. The Fetch Keywords from Sheet node uses Google Sheets API to retrieve rows from sheet ID 1319606837 in document 1jzDvszQoVDV-jrAunCXqTVsiDxXVLMGqQ1zGXwfy5eU.

Ensure you have the correct OAuth credentials set up for Google Sheets integration, as shown by the linked credentials node Google Sheets account NN DB test.

Once executed, you should see the list of keywords loaded into the workflow.

Common mistake: Incorrect document or sheet ID will cause data fetch to fail.

Step 3: Split Keywords into Batches for Processing

To efficiently process and prevent API overload, the Process Keywords in Batches node splits the fetched keywords into groups of 6. This batching improves workflow performance and compliance with rate limits.

Check the batch size parameter in the node settings to adjust as necessary, especially if you experience slow API responses.

Common mistake: Setting batch size too high causing API errors or too low leading to slow processing.

Step 4: Wait Node to Prevent Rate Limiting

The Prevent API Rate Limiting node introduces a pause in execution before sending data to the AI agent. This wait node helps avoid exceeding OpenAI API quotas and keeps the workflow stable.

The node currently uses default wait settings—adjust timing based on your API limits and error feedback.

Common mistake: Removing or minimizing the wait duration can trigger rate limit errors.

Step 5: Analyze Keywords with AI Agent Node

The AI Agent node powered by Langchain and connected to the OpenAI Chat Model evaluates each keyword batch. It uses a specific prompt:

Check the keyword I provided and define if this keyword has a name of the known IT software, service, tool or app as a part of it (for example, ServiceNow or Salesforce) and return yes or no.

The node extracts the {{ $json.Keyword }} to dynamically feed each keyword into the AI prompt.

The AI returns structured output parsed by the Structured Output Parser node as a simple JSON with “Isservice”: “yes” or “no.

Common mistake: Incorrect prompt text or missing JSON parsing will cause failures in keyword classification.

Step 6: Update Google Sheet with Results

Finally, the Update Sheet with Analysis Results node writes the classification back into the Google Sheet. It updates the “Service?” column matched by the keyword’s “Number” identifier.

This provides Anna a clear, up-to-date view of which keywords represent IT software/services, directly in her spreadsheet without manual review.

Common mistake: Incorrect column mapping or matching on wrong keys will overwrite unrelated data.

5. Customizations ✏️

• Change Batch Size: In the Process Keywords in Batches node, modify the batchSize parameter to process more or fewer keywords per round, balancing speed vs API limits.
• Adjust AI Prompt: Update the System Message in the AI Agent node to detect other keyword types like industries or competitor names.
• Expand Google Sheet Columns: Add new fields in the sheet and modify Update Sheet with Analysis Results node to save additional AI insights like confidence score or category tags.
• Add Error Handling: Insert error workflow or retry nodes after AI Agent to gracefully handle API failures and log issues.

6. Troubleshooting 🔧

Problem: “Error 429 Too Many Requests” during AI processing

Cause: The OpenAI API rate limit was exceeded due to too many rapid calls.

Solution: Increase the wait time in the Prevent API Rate Limiting node or reduce batch size in Process Keywords in Batches.

Problem: Google Sheets node fails to fetch or update data

Cause: Incorrect OAuth credentials or sheet/document ID mismatch.

Solution: Verify sheet ID and document ID in Fetch Keywords from Sheet and Update Sheet with Analysis Results nodes. Re-authenticate Google Sheets with correct permissions.

7. Pre-Production Checklist ✅

• Verify Google Sheets credentials have appropriate read/write permissions.
• Confirm the OpenAI API key is valid and connected.
• Run small test with manual trigger and 1-2 keywords to ensure data flows correctly.
• Check batch size and wait node timings match your API limits.
• Back up your Google Sheet data before mass updates to prevent accidental data loss.

8. Deployment Guide

Activate the workflow in n8n by enabling the trigger or scheduling it as needed. For manual runs, use the When clicking ‘Test workflow’ node.

Monitor execution via n8n’s Execution List for errors and logs. Since this workflow writes directly to your live Google Sheet, maintain backups regularly.

9. FAQs

Q1: Can I use this workflow with other AI models?
Yes, the OpenAI Chat Model node supports different GPT models; just update the model selection in its settings.

Q2: Does this consume many API credits?
The API usage depends on the number of keywords processed. Using batching and wait nodes helps optimize consumption.

Q3: Is the keyword data secure?
All data transmitted to the OpenAI API goes over secure HTTPS connections. Ensure you follow standard security practices in your n8n instance.

10. Conclusion

By deploying this n8n workflow, Anna now enjoys a fully automated system that quickly classifies keywords as IT software or service names using AI. This saves her several hours weekly, minimizes errors, and scales effortlessly with growing keyword lists.

Next, consider extending this setup to auto-generate content ideas for these keywords or integrating with marketing platforms like Salesforce or HubSpot for end-to-end campaign automation.

With the power of n8n and AI combined, your keyword research can go faster, smarter, and with far less hassle.