n8n Agentic Social Comment Analysis

Case Study – Agentic Mass Comment Analysis

In support of several different social media clients, I developed a simple yet powerful AI workflow designed to pull down social media comments and analyze them. Using Apify to run actors enables the workflow to bipass AI restrictions on social platforms. This scrapes all content and post information and then is formatted for analysis by an LLM. ChatGPT then reviews all comments and posts, classifying the content type, comment type, sentiment and topic. From here, all information is output into a Google Sheet that de-duplicates posts and adds recent ones. Using Gemini within sheets, we can then do ranking analysis and content evaluation by topic type, helping us understand which types of content yield the highest engagment, and positive engagement. Designed to run weekly, the credit cost is low since the focus is on limited channels. Easily expanded, this workflow has also been run on different social channels and even Reddit.

The Approach

Automation First

The process runs automatically every week. Scrapes, reviews, formats and exports into sheets without needing review.

Intelligence

Beyond saving incredible time, the process applies AI to categorize and evaluate content so strategists can focus on creative strategy.

Simplicity

n8n workflows can quickly spiral into complex messes. This is a minimal node, efficient approach to delivering excellence in a narrow scope.

Tools

Apify

I used Apify as an extremely low-cost application for scraping social content at scale, focused on selected social media accounts or channels.

ChatGPT

Several models would have served the role of categorizing, but I used a light-weight GPT LLM to classify content.

Gemini

Once inside a sheet, Gemini can readily be applied to content evaluation and data processing, providing graphs and performance indicators.

Impact

Time Savings

Performing a manual social media comment analysis took a strategist hours per account. This workflow easily saved 3-4 hours per week in comment review and analysis.

Commercialization

Several clients were not even aware of what AI tools could do in supporting strategy evaluation. This workflow added a new product that companies were ready to purchase.

Confidence

A simple, functioning agentic workflow was vital in getting team and client confidence. AI often seems fragile or unreliable, so getting an easy win helped grow confidence in our tools.

Learnings

Like many n8n workflows, it was easy to find myself spiraling on this workflow construction. Lots of nice to have features and extra functions that could be added, but each introduced more chances for the AI to slip in focus and hallucinated. By keeping the workflow clean and simple, execution has been near-flawless in quality and outputs reliable. When building a workflow to support existing work from a team, the output must be reliably precise and predictable, or adoption falls off. There are additional workflows I’ve built to draw of this and introduce content generation based on a strategic direction, but keeping this workflow simple was intentional and effective.