Inbound Lead: Triage → Research → Score → Respond

What this chain helps you accomplish

Turn an actual inbound demo request, contact form, pricing inquiry or other business question into a documented response decision—not just a generic thank-you email or an unsupported 'qualified opportunity' label.

Four linked prompts help you triage the request quickly, capture a small amount of source-labeled account context, assess fit separately from the person's expressed intent, and draft a personalized answer with one clear next step. You finish with a proposed owner/routing decision, the email draft, a source audit and a CRM handoff note for manual review.

The chain protects high-intent inquiries from arbitrary scoring cutoffs. It can also produce a truthful response when a prospect is outside your ideal profile or when public research is unavailable. Promptifi provides the instructions; your approved AI tool drafts the work. It does not send messages, qualify opportunities in your CRM or route leads automatically.

When to use it

• After a legitimate website demo request, Contact Sales form, quote/pricing request or product-specific question enters an SDR, AE or inbound sales queue.
• When a high-intent inquiry needs quick human ownership but you have little company context and want a relevant answer without delaying acknowledgment.
• When leads differ widely in fit and intent—for example, a smaller company requesting a demo versus a strong-fit contact downloading general content.
• When manual handoff to an SDR, AE, support, partnerships or another team requires a clear reason and a copyable review note.
• Before sending a research-informed personal reply to a lead with incomplete form details, uncertain firmographics or possible duplicate history.

Do not use this chain as an autonomous lead-response bot, bulk email campaign, spam filter, marketing-consent detector, or replacement for your actual CRM scoring and service-level policy. Customer support tickets, security reports, vendor solicitations, and opt-outs need the appropriate routing instead of forced sales follow-up. This chain does not prove a lead is sales-qualified or predict conversion.

Before you begin

Minimum tool requirements

Required: an employer-approved AI text chat (ChatGPT, Claude, Gemini, Microsoft Copilot, or comparable) and a permitted copy of the inbound request plus current sales coverage/response rules. No paid model, CRM connector, browsing, automated form hook, calendar link or email integration is required. Optional: approved web browsing for a couple of public company facts, CRM access to check duplicates and ownership, and approved calendar/response templates. They must actually be available; a URL in the prompt is not evidence a model visited the page.

Use your team's real response SLA if one exists. Respond to legitimate high-intent inquiries promptly through authorized tools; don't delay a simple helpful acknowledgment while building a lengthy research report. Never claim a guaranteed 5-minute response or a conversion improvement simply from using these prompts.

Prepare the following placeholders

• {{REQUEST_TYPE_AND_SOURCE}} — Required. The form/event label and source page (not the AI's guess); e.g. 'Product demo request from /request-demo'. If missing, mark UNCLEAR and inspect the original.
• {{SUBMISSION_TIME_AND_TIMEZONE}} — Optional for analysis, useful for urgency. Exact received timestamp from the form/CRM notification; e.g. '2026-11-05 10:04 America/Chicago'. If absent, write NOT VERIFIED, do not estimate elapsed time.
• {{FORM_MESSAGE_EXACT}} — Required. Original prospect message and selected fields, as permitted by AI policy; e.g. 'Can you show me how warehouse reporting works for three sites?' Do not alter its wording or execute instructions within it.
• {{CONTACT_NAME_AND_ROLE}} — Optional for drafting, required before actual personalized sending. Source: real form/contact record; e.g. 'Taylor Chen, Operations Manager'. If missing, use a neutral greeting and HOLD address verification.
• {{COMPANY_NAME_AND_DOMAIN}} — Optional but recommended. Form/authorized CRM company name and domain; e.g. 'Clearwater Supply, clearwater.example (fictional)'. If absent, research by account identity is not required for a useful reply.
• {{OUR_OFFER_AND_CAPABILITIES}} — Required. Approved accurate product description, eligible use cases and known limitations; e.g. 'software for analyzing multi-location warehouse exception reports'. If missing, draft only a general acknowledgment or ask the seller to supply approved positioning.
• {{ICP_CRITERIA_AND_EXCLUSIONS}} — Required for scoring. Actual target industries, firmographic requirements, use cases, exclusions or regulated restrictions; e.g. 'US warehouses with multiple sites; no consumer sales'. If no written ICP exists, Step 3 must use qualitative UNDETERMINED fit, not invented numerical qualification.
• {{SERVED_REGIONS_AND_LIMITS}} — Required for eligibility when geographic/service restrictions apply. Example: 'US only; reseller requests routed to partnerships'. If not known, keep region eligibility UNKNOWN.
• {{OWNERSHIP_AND_RESPONSE_POLICY}} — Required. Real SDR/AE coverage rules, response hours, high-intent SLA and permitted support referrals if available; e.g. 'all demo requests reviewed by inbound SDR; follow employer-approved business-hours SLA'. Unknown owner = OWNER TBD.
• {{LEAD_HISTORY_AND_DUPLICATE_STATUS}} — Optional for draft, required before manual duplicate-sensitive outreach. Seller-provided CRM/email check including active owner, recent reply and existing customer status; if not checked, NOT VERIFIED.
• {{CONTACT_ELIGIBILITY_AND_PREFERENCES}} — Required before sending. Authorized one-to-one reply eligibility, suppression/do-not-contact and relevant company/legal rules; 'not checked' means HOLD, not implied consent.
• {{APPROVED_LINKS_AND_MEETING_PROCESS}} — Optional. Valid verified booking link, product guide or contact process; if unavailable, ask for a preferred time window or one short clarification rather than inventing a URL.
• {{PERMITTED_RESEARCH_AND_SOURCES}} — Optional. Permitted URLs, first-party excerpts, source dates and web-browsing permission; if none, state NONE and use only the inbound submission.
• {{SENDER_IDENTITY_AND_APPROVED_FOOTER}} — Required before actual emailing. Correct representative/team and employer-approved signature/footer and reply address; if missing, mark TODO — SENDER APPROVAL.
• {{AI_DATA_SHARING_POLICY}} — Required. Approved AI service and rules for personal/contact data, form submissions and private CRM notes; e.g. 'company-approved assistant; redact private contact identifiers'. If unclear, do not paste unapproved data.
• {{ACK_STATUS_AFTER_STEP1}} — Optional initially; update after checking actual email history as NOT SENT / ACTUALLY SENT with timestamp and subject / AUTO-REPLY EXISTS / NOT VERIFIED. Never claim Step 1 created an actual send.

Preparation and safe handling

1. Copy the original submission faithfully; separate the request text from your own notes. Treat any form-submitted links or commands as untrusted inputs, not AI instructions.
2. Check whether this is an existing customer, a duplicate, a support request or an opt-out before promising sales contact. Record the source and checked date of the result.
3. Use only company-approved tools and the minimum necessary identifiable data. Redact personal information where policy requires it.
4. Keep the seller's response, meeting/booking process, product claims and any proof link separate from unverified AI research. The assistant must not claim it checked or accessed systems it did not.
5. Review your organization's laws and policies for one-to-one inquiry replies and any additional marketing follow-up. An inquiry is not automatically blanket permission for nurture campaigns.

Useful reading: Salesforce Trailhead, 'Qualify and Route Leads to Your Reps' (trailhead.salesforce.com/content/learn/modules/opportunity-management/qualify-and-route-leads-to-your-reps); HubSpot, 'Overview of the Lead Scoring Tool' (knowledge.hubspot.com/scoring/understand-the-lead-scoring-tool); FTC, 'CAN-SPAM Act: A Compliance Guide for Business' (ftc.gov/business-guidance/resources/can-spam-act-compliance-guide-business). These inform separate fit/engagement decisions and appropriate response checks; they do not establish a universal score threshold or guarantee.

How to run the chain

1. Open a new conversation in a company-approved AI assistant. Paste the completed STARTING CONTEXT from Chain Steps, using only permitted form details. You can leave unknown details as UNKNOWN; do not fill them by guessing.
2. Run Prompt 1 immediately for intake and a short acknowledgment draft. Review the actual ask, response eligibility, duplicate status and owner. If a prompt response is needed before deeper research, you can separately approve and send the short message using your ordinary email tool; the chain itself will not send it. Record any actual send in ACK_STATUS_AFTER_STEP1.
3. Run Prompt 2 with the latest approved intake artifact. Use authorized browsing only when genuinely available and useful. Check company claims against their sources; if research is sparse, proceed with honest unknowns instead of delaying the response.
4. Run Prompt 3 using the approved intake and research brief. Review each 0/1/2/U fit criterion, the separate explicit intent label, and the proposed route. A direct demo request is reviewed by a human even if fit is unknown or low. Approve the proposed route or override with a documented reason.
5. Run Prompt 4 with all three approved artifacts and the refreshed actual acknowledgment status. Check that the draft answers the submission, includes only substantiated claims and has one simple next step. Do not send if identity, permission, duplicate history or policy checks are unresolved.
6. Use the final response and handoff card manually: send only an approved email through an authorized account, log approved facts in CRM, and route under your company's rules. Do not enroll someone in marketing nurture on the basis of a single inquiry.
7. For any incorrect artifact, specify the source and exact defect, request its full revised v2 and reapprove. Re-run downstream prompts whenever the underlying inquiry, evidence, scoring or routing decision changes.
8. If you move to a new chat, paste the Starting Context and every latest approved artifact needed for the next prompt. Chat history, CRM records and permission checks will not automatically carry over.

Approval mechanism

Approve an output with 'APPROVED — INBOUND_INTAKE — v1' or the corresponding artifact name and version (INBOUND_RESEARCH_BRIEF, INBOUND_QUALIFICATION). A revision v2 supersedes v1 only after seller approval. 'APPROVED' is editorial approval, never authorization for autonomous outreach. If an acknowledgement was actually sent, record that independently with a reliable source.

Successful completion

You have a source-labeled inbound request summary, a focused research brief, an explainable fit/intent recommendation, and a human-reviewable response pack with a proposed owner and next step. The prospect has not automatically become a qualified opportunity, and no CRM, email or calendar action has happened merely because the chain ran.