Mine YouTube Objections in n8n: Bright Data + OpenAI + Slack

Mine YouTube Objections in n8n—Bright Data + OpenAI + Slack

Turn competitor YouTube comment sections into actionable sales insights. This n8n workflow crawls review videos via Bright Data, classifies every comment into objection types, and uses OpenAI to generate a crisp digest—then posts the results to Slack and logs them in Google Sheets.

What this workflow does

• Starts on schedule: runs every Monday morning (or can be started manually) to launch a new research run.
• Crawls competitor review comments: normalizes YouTube video URLs/IDs from workflow settings and triggers a Bright Data dataset crawl to collect top or newest comments per video—without needing a YouTube API key or dealing with YouTube quota.
• Waits for results + handles gaps: polls until the crawl is ready, downloads the snapshot results (including per-video errors), then filters/deduplicates rows (skipping very short comments).
• Classifies objections + extracts rivals: sorts each comment into one or more themes: price, performance, complexity, missing features, switching, and trust. It also extracts named alternatives when viewers mention switching.
• Creates an OpenAI digest: ranks themes by volume and likes, selects representative quotes, and generates one-sentence insights per theme plus an overall summary.
• Publishes results: appends one row per theme to Google Sheets and posts the run digest (top quotes, named alternatives, and any crawl gaps) to a Slack channel.

Use cases

• SaaS operators tracking what prospects dislike about competitor products after YouTube review uploads.
• Automation engineers building a repeatable “voice of customer” pipeline from YouTube comments.
• Product teams using objections (price, performance, complexity, missing features, switching, trust) to guide messaging and roadmap priorities.

Technical details

• Integrations: Bright Data (dataset crawl with Authorization: Bearer <YOUR_API_KEY>), OpenAI, Slack, and Google Sheets.
• n8n nodes / logic: if, set, code, wait, slack, split (with code-based normalization, filtering/deduping, and theme assignment).