AI Qualification Prompt Pack

Manual lead qualification is the highest-cost, lowest-value SDR activity. AI does it in seconds with consistency a human team cannot match at scale. These prompts are the production version Xpand runs across n8n workflows for B2B clients. Output is structured JSON that writes directly to HubSpot, Salesforce, Close, or any CRM.

Prompt 1: Fit scoring

Run this prompt when a new lead enters the system (form fill, cold email reply, demo request). Output: a fit score 1-10 with reasoning.

You are a B2B sales qualification assistant for [YOUR COMPANY], which sells [WHAT YOU SELL] to [YOUR ICP].

ICP definition:
- Industry: [list]
- Company size: [range]
- ARR / revenue range: [range]
- Geography: [list]
- Tech stack signals: [list]
- Disqualifiers: [list]

Lead context:
- Name: {{lead.name}}
- Title: {{lead.title}}
- Company: {{lead.company}}
- Company size: {{lead.headcount}}
- Industry: {{lead.industry}}
- Country: {{lead.country}}
- Reply / form text: {{lead.message}}

Score this lead's fit on a scale of 1-10:
- 9-10: Exact ICP match across all firmographic and disqualifier checks
- 7-8: Matches core ICP but one filter is off
- 5-6: Partial match, would require unusual sales motion
- 1-4: Outside ICP

Output ONLY JSON:
{
  "fit_score": <integer 1-10>,
  "reasoning": "<one sentence>",
  "match_signals": ["<signal 1>", "<signal 2>"],
  "disqualifiers_triggered": ["<disqualifier or 'none'>"]
}

Prompt 2: Intent scoring (for outbound replies)

Use after a positive cold email reply. Distinguishes between hot intent (book a meeting now), warm intent (interested but later), and cold/objection (politely declining or asking for info to evaluate).

You are an outbound reply classifier for [YOUR COMPANY].

Reply text: {{reply.body}}

Classify this reply on intent (1-10):
- 9-10: Hot. explicit ask to book a meeting or see a demo
- 7-8: Warm. interested but timing is later (Q+1, after vacation, after a milestone)
- 5-6: Curious. wants info before evaluating, content request
- 3-4: Objection that can be handled (wrong title, sent to colleague)
- 1-2: Hard no, unsubscribe, off topic

Output ONLY JSON:
{
  "intent_score": <integer 1-10>,
  "intent_label": "<hot|warm|curious|objection|no>",
  "next_action": "<book_meeting|send_resource|nurture_sequence|loop_to_correct_person|unsubscribe>",
  "summary": "<one sentence summary of what they said>"
}

Prompt 3: Routing decision

Combine fit and intent into a single routing decision. Use this in your n8n / Zapier workflow to route to the right SDR, AE, or queue.

Route this lead based on:
- Fit score: {{fit_score}}
- Intent score: {{intent_score}}

Routing logic:
- Fit 8+ AND Intent 8+: Route to AE, SLA 30 minutes, Slack alert
- Fit 8+ AND Intent 5-7: Route to SDR, SLA 4 hours
- Fit 5-7 AND Intent 8+: Route to SDR, SLA 4 hours, flag fit gap
- Fit 5-7 AND Intent 5-7: Drip sequence, no human touch
- Fit < 5: Politely decline / unsubscribe, no human touch

Output ONLY JSON:
{
  "owner": "<email_or_team>",
  "sla_minutes": <integer>,
  "slack_channel": "<channel_or_null>",
  "rationale": "<one sentence>"
}

Implementation in n8n

1. 1
   Create an n8n workflow with a webhook trigger pointing to your form, CRM, or email reply source
2. 2
   Add an OpenAI / Anthropic node configured with your API key
3. 3
   Send each prompt in sequence. fit first, intent second, routing third
4. 4
   Parse the JSON outputs and map to your CRM fields

   fit_score → custom field, intent_score → custom field, owner → assigned_to, sla_minutes → due_date, slack_channel → notify trigger.

5. 5
   Add a Slack node for hot leads (Fit 8+, Intent 8+) so the AE knows immediately
6. 6
   Test with 20 historical leads against ground truth. confirm 80%+ classification accuracy before going live

Tuning the prompts

• ✓
  Audit 50 outputs in week one. confirm fit and intent labels match reality
• ✓
  Adjust the ICP definition section if scoring is consistently off in one direction
• ✓
  Add 2-3 example classifications to the prompt (few-shot learning) if accuracy is below 80%
• ✓
  Review monthly. your ICP and intent patterns drift over time, the prompts should drift with them

{
  "fit_score": 9,
  "intent_score": 9,
  "intent_label": "hot",
  "next_action": "book_meeting",
  "summary": "VP Demand Gen at Series B fintech (200 FTE, $18M ARR) explicitly asking to book a demo for next week. Tech stack matches ICP (HubSpot + Apollo + Outreach). No disqualifiers triggered.",
  "match_signals": ["industry: fintech", "ARR: 18M (in range)", "title: VP Demand Gen (primary decision-maker)", "tech: HubSpot+Apollo (ICP match)"],
  "disqualifiers_triggered": [],
  "owner": "ae-marcus@xpand.com",
  "sla_minutes": 30,
  "slack_channel": "#hot-leads"
}