n8n Tutorial: YouTube Comment Buy Signals → HubSpot Lead + Auto-Reply Automation

You post a YouTube video, and within hours the comments fill up. Most are noise — but buried in there are the ones that matter: "Where do I buy this?", "Does this work with HubSpot?", "How much for the pro version?". Those are buyers raising their hand. And by the time you scroll back to find them — if you ever do — they've cooled off, bought elsewhere, or forgotten you. Manual comment triage doesn't scale past your first viral video, and it fails at exactly the moment you need it most.

The Problem: Buy Signals Die in the Comment Section

YouTube comments are one of the highest-intent, lowest-friction channels a technical creator or SaaS founder has. Someone who watched a 12-minute tutorial and then typed a question is far warmer than a cold ad click. But three things kill that intent:

• Latency. A reply 48 hours later reads as automated indifference. A reply in the first two hours reads as attentiveness.
• Volume. Once you have more than a handful of active videos, no human is reading every comment, let alone classifying intent.
• No system of record. Even if you spot a buyer, they never make it into your CRM. There's no follow-up sequence, no attribution, no pipeline. The signal evaporates.

The result: your best inbound leads are sitting in public, unqualified and unrouted, while you pay for ads to acquire colder ones.

The Solution: A Comment-to-CRM Loop That Runs Every 2 Hours

The fix is a scheduled n8n workflow that treats your comment section like an inbox. Every two hours it pulls fresh comments across your channel, classifies each one for purchase intent, and then acts on the buyers automatically: it posts a contextual reply and writes the person into HubSpot as a lead with the comment, video, and detected intent attached.

The logic in plain terms:

1. Fetch new comments since the last run.
2. Score each comment for buy intent (keywords + optional LLM classification).
3. For high-intent comments: auto-reply with a helpful, non-spammy message pointing them to the next step.
4. Upsert the commenter into HubSpot with intent, source video, and the raw comment text.
5. Record the comment ID so it's never processed twice.

It runs unattended, it's deterministic, and it turns a public comment thread into a qualified pipeline entry — without you touching YouTube Studio.

Step-by-Step: Building It in n8n

Here's the node-by-node structure. If you're building from scratch, this is the skeleton; the ready-made template ships with all of it wired and de-duplicated.

1. Schedule Trigger

Start with a Schedule Trigger node set to an interval of 2 hours. Two hours is the sweet spot: fast enough that replies still feel live, slow enough to stay well under YouTube API quota. Avoid a 1-minute cron here — you'll burn quota for nothing.

2. Fetch Comments (HTTP Request or YouTube node)

Use an HTTP Request node against the YouTube Data API v3 commentThreads endpoint (part=snippet, allThreadsRelatedToChannelId, order=time), authenticated with a Google OAuth2 credential scoped to youtube.force-ssl. Pull the most recent 50–100 threads. This one call covers your whole channel, so you don't loop per-video.

3. De-duplicate Against State

Add a Code node (or a Filter node backed by n8n static data / a small datastore) that drops any comment whose id you've already handled. Store processed IDs in workflow static data or an external store. This is the single most important step — skip it and you'll re-reply to the same person every two hours and get flagged as spam.

4. Classify Buy Intent

Two-tier approach. First a fast Filter / IF node matching high-signal phrases: buy, price, cost, purchase, how much, where do I get, discount, demo, trial. For higher precision, route survivors through an AI/LLM node (or an HTTP call to Claude) with a prompt that returns a strict JSON label like {"intent":"buyer|question|noise","confidence":0-1}. Only buyer with confidence above your threshold proceeds. The keyword pre-filter keeps your LLM costs near zero because 90% of comments never reach it.

5. Auto-Reply

For confirmed buyers, an HTTP Request node POSTs to the YouTube comments endpoint (part=snippet) with a reply parented to the original comment thread. Keep the reply human and specific — reference the video topic and give one clear next step and link. Rotate 3–4 message variants (a Set node picking by index) so replies don't look copy-pasted.

6. Upsert to HubSpot

Finish with the native HubSpot node (Contact → Create/Update). Map the commenter's display name and a synthesized identifier into the contact, and push the comment text, source video URL, detected intent, and confidence into custom properties (e.g. yt_comment, yt_video, lead_intent). Set the lifecycle stage to lead. Now your existing HubSpot workflows — sequences, tasks, Slack alerts — can fire off a real inbound signal.

Benefits: What Changes Once This Runs

• Reply latency drops to under two hours, automatically, across every video you've ever published — including old ones that keep getting traffic.
• Buyers land in your CRM with full context, so follow-up is a HubSpot sequence, not a manual scavenger hunt.
• Attribution finally works: you can see which videos generate purchase-intent comments and double down on that content.
• Zero marginal effort. A channel doing 500 comments a week costs the same operator time as one doing 5 — none.
• Your comment section becomes a sales asset instead of a support liability.

Common Pitfalls (and How to Avoid Them)

• Skipping de-duplication. The number one failure. Without a processed-ID store, every run re-replies to old comments. Persist state before you turn on the schedule.
• Over-triggering the auto-reply. A loose keyword filter will reply to "this is too expensive" as if it were a buyer. Tune your threshold and let the LLM tier arbitrate ambiguous cases.
• YouTube quota exhaustion. The Data API has a daily quota, and writes (posting replies) cost more units than reads. Batch your fetch, cap replies per run, and keep the interval at 2 hours — not minutes.
• Robotic replies. Identical replies at scale get pattern-flagged and annoy real humans. Rotate variants and reference the specific video.
• OAuth token expiry. Google OAuth2 refresh tokens can go stale if the credential is revoked or unused. Add an error-workflow branch that alerts you (Slack/email) on a 401 so a dead auth doesn't silently kill the pipeline.
• Missing HubSpot custom properties. The upsert fails silently if lead_intent or yt_video don't exist. Create the properties in HubSpot first, then map them.

Build the de-dup and the intent threshold carefully and the rest is mechanical. Once it's live, you stop losing your warmest leads to the scroll — every buy signal gets a fast reply and a permanent home in your CRM, on autopilot, every two hours.

PRONTO PARA USAR

Ja construimos isso pra voce

Nao comece do zero. O YouTube Comment Buy Signals → HubSpot Lead + Auto-Reply e um workflow n8n pronto para instalar — conecta suas ferramentas em minutos, sem codigo.

Instalar por $79 →