Two hundred badge scans, filed and drafted inside 48 hours

What is the exact prompt?

Feed it the scanner export the evening the booth comes down:

• "Here is the badge-scan CSV from [event]. First, clean it: dedupe on email, drop rows with no company, and split the remainder into three tiers by the notes column: CONVERSATION (a note describing what we discussed), SCAN (a note like 'stopped by' or none), COMPETITOR/VENDOR (their company sells to us or against us)."
• "Enrich every CONVERSATION and SCAN row via Apollo: title, company size, and whether the company is already in our CRM. Create or update CRM records, tagging everything [event]-[year]. Where the company already has an open deal, add a note to the deal instead of creating a duplicate."
• "For each CONVERSATION row, draft an email that leads with the note: reference the specific thing we discussed, answer any question the note says they asked, and propose one concrete next step. For SCAN rows, draft one short two-sentence template referencing the event only. Different subject lines per tier."
• "Leave every draft unsent. Show me the CONVERSATION drafts first, worst note to best, so I can fix the weak ones."

The notes column decides everything

The note is the only thing separating a lead from a scan.

Treating them identically is how a booth sends "great to connect!" to someone who asked a specific integration question.

The tier split puts effort where it returns: forty personal drafts you review, one honest template for the polite scans. Reviewing worst note first catches the padding ("loved chatting about your data challenges") while you are still paying attention.

What does enrichment add that the badge did not?

Routing. The badge says who they are; Apollo says whether they can buy. Title and size decide which follow-ups come from a rep, and the already-in-CRM check catches cold-emailing an existing customer or duplicating a live deal. Same discipline as lead enrichment, pointed at a CSV.

Tagging every record with the event name makes "did that conference pay for itself" a CRM filter at pipeline-review time rather than a debate at budget time.

Every send still goes through you

Forty drafts built on real notes take twenty minutes to review, and the SCAN template goes out on one decision instead of eighty. The Gmail gate is connector policy rather than prompt text, and Ask Always in Settings, AI permissions widens it to everything; the security model covers both.

The warmest booth conversations deserve more than an email. Conference follow-up and prospect research are the next plays in the chain.

Frequently asked questions