Operations

Intent Classification Prompt for Inbound Email

Sales, Support, Partnership, Press, Spam — route inbound email by intent, with a Strict "Other" so the weird ones reach a human.

Overview

Intent routing is where wrong labels hurt: a press inquiry routed to sales support dies quietly. This setup classifies inbound email into five intents under Strict ambiguity — if nothing clearly fits, the model returns "Other" instead of forcing the nearest label, because in routing, a forced wrong label is silent and an "Other" is visible. Definitions carry the borders (Support Request is "an existing customer asking for help" — prospects asking questions are Sales Inquiry), and high/medium/low confidence enables auto-routing the clear cases.

Workflow

Classify before any human reads
The prompt runs on the raw email; quoted history and signatures don't derail a label decision the way they derail extraction.
Let Other be the safety lane
Everything the model can't confidently place lands in one reviewable bucket instead of five wrong ones.
Tune the Support/Sales border
"Existing customer" is the deciding phrase — if your prospects also ask product questions, the definitions are where you encode that.

Why This Works

Strict + Other converts silent misroutes into visible review items
Customer-status-based definitions resolve the commonest routing confusion
Confidence output lets you automate the clear 80% and human-review the rest

Best for

Inboxes where misrouted mail dies quietly
Teams that want auto-routing with a human fallback lane
Intent sets where "existing customer vs prospect" decides the route

Not for

Extracting the sender's details and request — that's the Extraction Prompt Generator
Drafting replies — routing decides the desk, not the answer

Use cases

Routing a shared inbox to sales, support, and partnerships
Catching press inquiries before they rot in the wrong queue
Filtering spam without a separate spam model

Customize This Resource

Opens this setup in Data Classification Prompt. Generate to get the full classification prompt — then adjust the labels, ambiguity mode, and confidence output.

Open in Data Classification Prompt

Prompt Template

TASK
Classify the intent of inbound emails for routing to the right team.
Assign exactly one of the labels defined below to the text in the input.

LABELS
- Sales Inquiry: Interest in purchasing, upgrading, or pricing details.
- Support Request: An existing customer asking for help with the product.
- Partnership: A proposal to collaborate, integrate, or resell.
- Press: A journalist or media request for comment or information.
- Spam: Unsolicited, promotional, or malicious content with no genuine intent.
- Other: Does not clearly fit any label above.

CLASSIFICATION RULES
- Read the entire text before deciding — the deciding signal may come late.
- Judge by content and intent, not by tone, length, or formatting.
- Match the text against the definitions, not just the label names.
- Use only the labels defined above, plus "Other" when nothing fits — never invent any other label.

EDGE CASE RULES
- If the text fits more than one label, choose the one matching its primary purpose — what the writer mainly wants.
- If two labels fit equally, prefer the more specific definition over the more general one.

AMBIGUITY POLICY
- If no label clearly fits, return exactly "Other" — do not force a fit.
- "Other" is a valid answer; a forced wrong label is not.

CONFIDENCE
- With each label, report a confidence level: high, medium, or low.
- high = the text unambiguously fits the definition; medium = a good fit with minor uncertainty; low = the best available choice among the labels.
- Confidence reflects your certainty about the fit — it is self-reported, not computed.

OUTPUT FORMAT
Return your classification as a single valid JSON object with a "label" key and a "confidence" key.

EXAMPLE OF A VALID RESPONSE
{
  "label": "Sales Inquiry",
  "confidence": "high"
}

OUTPUT RULES
- Return only the JSON object — no text before or after it.
- Do not wrap the output in markdown code fences.
- Return the label text exactly as defined above — no paraphrasing, no abbreviations, no new labels.

Related Resources

Explore more resources that work well with this one.