Your comments already contain buyers. The fastest way to turn them into leads is a 14-day sprint: three days to find the buying signals you've been missing, four to set up rules and AI-drafted replies, then a week to run it and count the results. The goal isn't to automate everything. It's to make sure every comment that asks "how much?" gets a quick answer and an owner.
TLDR: Audit 30 days of comments for buying questions (Days 1 to 3), write keyword rules and reply drafts with approval switched on (Days 4 to 7), run the hand-off to sales for a week (Days 8 to 12), then count leads and tune the rules (Days 13 and 14).
Days 1 to 3: find the buying comments you already have
Export or scroll back through 30 days of comments and DMs on your three most active posts or ads. Mark every comment that asks something a buyer would ask.
For each one, note the exact words, the post, how long it waited, and whether anyone answered. You're building a list of lead signals in your customers' own phrasing. Typical ones:
- "Price?", "How much", "Cost for 10 people"
- "Link", "Where can I buy", "Do you ship to..."
- "Is this available in...", "Can I book", "DM me details"
On day 3, count: how many were buying questions, how many got a reply, and the median wait. That count is your baseline. If you want a faster version of this audit, the 5-minute reply audit is a good warm-up.
Days 4 to 7: set up rules, drafts and approval
Now turn the signal list into rules. Each rule has three parts: the words that trigger it, the action, and who approves.
| Signal | Action | Approval |
|---|---|---|
| "Price", "cost", "how much" | Tag as Lead, draft a reply that points to pricing | Manual at first |
| "Link", "guide", "send me" | Reply under the comment and send the link by DM | Auto, after a test |
| "Book", "demo", "quote" | Tag as Lead and assign to a salesperson | Manual |
| Complaint wording | Tag as Risk, notify the owner, no auto reply | Manual |
In Mydrop's social inbox keyword rules can match with exclusions, then tag, assign, notify or draft an AI reply. Approval can be manual, automatic or hybrid per rule. For giveaways and lead magnets, trigger words reply under the comment and DM the link or file to anyone who comments your keyword.
Write the reply drafts in your own voice first, then let AI vary them. Two rules keep it honest: every reply answers the actual question, and no reply promises something the offer page does not.
Days 8 to 12: run it and hand leads to a person
For five days, keep every rule on manual approval and read each draft before it goes out. You're checking two things: does the keyword catch what it should, and does it catch what it shouldn't?
Fix as you go. If "price" also catches "price match complaints", add an exclusion. If a draft sounds like a brochure, rewrite the template.
Agree on one hand-off rule with whoever sells. For example: a Lead-tagged thread gets a human reply within four working hours, and the thread owner logs the outcome as won, lost or no answer. Without that last part, you can't count anything on day 14.
Days 13 and 14: count leads and decide what to automate
Pull the tagged threads and answer four questions:
- How many Lead-tagged threads were there, compared with the baseline you counted on day 3?
- What was the median time to first reply now?
- How many turned into a conversation, a booking or a sale?
- Which rules produced noise?
Switch to auto-approve only the rules that had no bad drafts for five days, and keep money, legal and complaint topics on manual for good. Export the threads to CSV if finance or the sales lead wants to see them.
Be careful not to claim credit for sales you can't trace. If a comment led to a DM and then a call, log the source in your CRM; if you can't link them, count the conversations and say so.
Start the audit today, with 30 comments
You don't need the whole system to see the leak. Open your three most recent posts and mark every buying comment you find in ten minutes. If the list is longer than you expected, start free with Mydrop, connect the profile, and build your first three rules from that list.
Google review
Trustpilot review
Product Hunt comment
