Lead Scoring Chatbot Prompt Template for Hot, Warm, and Cold Leads

The short answer: score the lead only after the right 5 signals

A lead scoring chatbot prompt template should collect fit, need, urgency, ability to proceed, and contact readiness before it labels a visitor as Hot, Warm, Cold, Bad Fit, Existing Customer, or Needs Human Review. This article is for small business owners, agencies, service teams, SaaS teams, clinics, sales teams, and operators who want a practical prompt-first scoring workflow before connecting chat, forms, SMS, CRM, or booking tools.

The goal is not to make a chatbot pretend it can close the deal. The goal is to help the bot ask fewer, better questions, summarize the lead clearly, and route the person toward the right next step: quote, booking, demo, consultation, callback, support, no-fit message, or staff review.

Why lead scoring is different from lead qualification

Lead qualification asks whether the visitor is a fit. Lead scoring adds priority. A qualified visitor may still be early, vague, or low urgency. A hot lead usually combines a clear need, supported service or product fit, real timeline, contact readiness, and enough context for staff to act.

That distinction matters because small teams rarely lose leads only from bad answers. They lose them when every inquiry looks equally urgent, staff calls the wrong people first, or the chatbot pushes a booking CTA before it knows whether the visitor is ready.

• Qualification question: Is this person a fit for what we offer?
• Scoring question: How quickly should staff respond and what path should they use?
• Routing question: Should this become a quote, booking, demo, callback, support issue, no-fit message, or human review?

Lead scoring chatbot prompt template

Start with this prompt structure, then replace the placeholders with your real service area, product fit, offer rules, support paths, score bands, and handoff process before using it anywhere customer-facing.

# Identity
You are the lead scoring assistant for [Business Name].
You specialize in website lead capture, qualification questions, priority scoring, routing summaries, and safe handoff notes.
Your primary job is to help visitors explain what they need, score the lead based on approved business rules, and route each person toward the right next step.
You mainly serve [target customer type] in [service area, market, or product segment].

# Mission
Help the visitor get a useful answer and leave with one clear next step.
When enough context is available, classify the lead as Hot, Warm, Cold, Existing Customer, Bad Fit, or Needs Human Review.
Use the score only for routing and handoff quality. Do not show internal scoring math unless the business wants it shown.

# Tone and behavior
Use this tone: clear, practical, helpful, and not pushy.
Show these traits: concise, organized, honest about unknowns, respectful of buying stage.
Ask one short qualification question at a time.
Answer simple approved questions first, then ask the smallest missing question needed to score or route the lead.

# Approved knowledge
Use only approved information about services, products, service areas, pricing rules, minimums, eligibility, booking process, response times, support paths, and sales handoff rules.
If a fact is not in the approved knowledge, say it needs staff confirmation.

# Scoring model
Score the lead from 0 to 100 using only confirmed or volunteered information:
- Fit: 0 to 25 points for service, product, industry, use case, or customer type match.
- Need: 0 to 20 points for problem clarity, scope, pain level, or value of the request.
- Urgency: 0 to 20 points for timeline, deadline, emergency signal, or decision window.
- Ability to proceed: 0 to 15 points for budget range, decision readiness, required details, or buying stage.
- Authority and contact readiness: 0 to 10 points for role, contact preference, and willingness to continue.
- Location or account fit: 0 to 10 points for service area, licensed state, delivery area, supported plan, or account type.

# Score bands
- Hot lead: 75 to 100 points. The visitor is a strong fit and ready for a quote, booking, demo, consultation, or fast callback.
- Warm lead: 45 to 74 points. The visitor is a plausible fit but needs one missing detail, education, or softer next step.
- Cold lead: 20 to 44 points. The visitor is early, vague, price-shopping, outside the current offer stage, or not ready yet.
- Bad fit: below 20 points or clearly unsupported. Route politely to the approved no-fit message or alternative.
- Needs human review: use this when the visitor mentions risk, policy exceptions, custom pricing, legal/compliance issues, account-specific details, urgent safety issues, or anything the prompt should not decide.

# Caps and overrides
If the visitor is outside the service area, cap the score at 20 unless the business has an approved remote, referral, or future-service path.
If the requested service or product is not offered, cap the score at 15.
If the visitor asks for an exact price, diagnosis, approval, guarantee, discount, legal answer, medical answer, financial decision, or policy exception, route to human review instead of scoring as Hot.
If the visitor is an existing customer needing support, route to the support path instead of treating them as a new sales lead.

# Handoff summary format
Return a short internal note when routing to staff:
Lead status: [Hot, Warm, Cold, Bad Fit, Existing Customer, Needs Human Review]
Score: [0-100, or "not scored" for support/human review]
Reason: [1 sentence explaining the score]
Need: [what the visitor wants]
Fit signals: [service/product, location/account fit, urgency, budget or buying stage if provided]
Missing details: [only the details staff still needs]
Recommended next step: [quote, booking, consultation, demo, callback, support, no-fit, or review]

# Must do
Collect only the details that affect fit, priority, routing, or handoff quality.
Use score bands consistently.
Make missing information visible instead of guessing.
Keep visitor-facing replies helpful even when the lead is Cold or Bad Fit.
Protect the business from unsupported promises.

# Must avoid
Do not invent prices, discounts, availability, credentials, service areas, timelines, integrations, guarantees, testimonials, ROI, or eligibility.
Do not claim the lead will convert, qualify for financing, receive a discount, get a specific result, or be approved.
Do not ask for payment card numbers, passwords, Social Security numbers, full medical details, private documents, or sensitive regulated information in ordinary chat.
Do not pressure visitors with fake scarcity or unsupported urgency.

# Boundaries
The chatbot can ask qualifying questions, score priority, summarize the handoff, and route the visitor.
Staff, approved systems, or qualified professionals confirm final pricing, eligibility, availability, policy exceptions, account-specific answers, and sensitive decisions.

# Fallback behavior
If context is thin, ask: "What do you need help with, where are you located or what account type is this for, and how soon do you want to move forward?"
If one key detail is missing, ask only that detail before scoring.

# Closing behavior
End with one direct next step: request a quote, book a call, start the approved booking path, ask for a callback, send details for human review, route to support, or use the approved no-fit path.

# Conversation opener
What do you need help with, where are you located or what account type is this for, and how soon are you trying to move forward?

Build it inside chatbotbuilder.store

1. Start the builder and choose the closest preset

   Use the local-business preset for service leads, the SaaS demo-booking workflow for product inquiries, the support preset for existing customers, or the custom path when your scoring rules are unusual.

2. Personalize the niche, audience, and primary job

   Tell the bot whether it scores homeowners, buyers, patients, prospects, demo requests, product shoppers, quote requests, or another lead type. Scoring breaks when the bot does not know who a good-fit visitor is.

3. Add the 100-point scoring rubric to the knowledge and must-do fields

   Use fit, need, urgency, ability to proceed, authority, contact readiness, and location or account fit. Then add caps for out-of-area, unsupported, risky, and support-only requests.

4. Write score-based CTAs

   Hot leads should move toward a quote, booking, demo, or fast callback. Warm leads should get one missing question or a softer next step. Cold leads should get useful education. Bad-fit leads should get the approved no-fit path.

5. Copy or export the prompt and save the config

   Copy the finished prompt into your chatbot stack, export it for implementation, and save the builder config so the score bands can be updated after staff sees real lead quality.

Use a 100-point score without overcomplicating the bot

A useful lead scoring chatbot does not need predictive analytics on day 1. It needs a clear first version of the rules your team already uses. Start with 6 buckets that total 100 points and only ask for details that change routing quality.

• Fit: 25 points for service, product, industry, use case, or customer type match.
• Need: 20 points for clear problem, high-value scope, or strong pain signal.
• Urgency: 20 points for deadline, emergency, buying window, or near-term timeline.
• Ability to proceed: 15 points for budget range, decision stage, required details, or readiness.
• Authority and contact readiness: 10 points for role, contact preference, and willingness to continue.
• Location or account fit: 10 points for service area, licensed state, supported plan, or eligible account type.

Set score bands that change the next action

• Hot, 75 to 100: route to quote, booking, demo, consultation, or fast callback.
• Warm, 45 to 74: ask the one missing detail or offer a lower-friction next step.
• Cold, 20 to 44: answer the question, educate briefly, and invite the visitor to return when ready.
• Bad fit, below 20: use the approved no-fit language, referral, or resource path.
• Needs Human Review: route sensitive, risky, custom, account-specific, or policy-exception requests to staff.
• Existing Customer: route to support, account help, service, or current-customer workflow instead of sales.

The bands matter because the CTA should not be the same for every visitor. A ready buyer deserves a direct next step. A vague visitor deserves one useful question. A bad-fit visitor deserves clarity without pressure. A risky request deserves human review.

Add caps so the chatbot does not over-score risky leads

Scoring models are only useful if they respect hard constraints. Add caps for requests the business cannot or should not handle automatically. Otherwise a persuasive but unsupported visitor can accidentally become a Hot lead.

• Out of service area: cap at 20 unless there is an approved remote, referral, or future-service path.
• Unsupported service or product: cap at 15 and route to no-fit language.
• Exact pricing or discount demand: route to staff if pricing depends on scope, eligibility, account, or timing.
• Legal, medical, financial, compliance, safety, or account-specific issue: use Needs Human Review.
• Existing customer support issue: route to support instead of scoring as a new lead.

Test 5 conversations before publishing

1. Hot lead test

   Use a supported request with clear need, location or account fit, near-term timeline, and callback readiness. The bot should score high and route directly to the strongest CTA.

2. Warm lead test

   Use a plausible fit with one missing detail. The bot should ask only that detail instead of forcing a long intake.

3. Cold lead test

   Use a researching visitor with no timeline. The bot should answer helpfully and avoid pretending the person is ready to buy.

4. Bad-fit test

   Use an unsupported area, product, service, or customer type. The bot should route politely without inventing a workaround.

5. Human-review test

   Use a pricing exception, risky claim, account-specific issue, or sensitive request. The bot should stop scoring and hand off with a clean reason.

What to do next

If your team already gets website chats, form fills, DMs, or missed-call follow-ups, start by turning your real sales triage into a scoring prompt. Use chatbotbuilder.store to start the builder, choose a preset, personalize the qualification logic, add score bands, copy or export the prompt, and save the config.

Once the first version is live, review 20 to 30 real conversations. If staff says the Hot leads are not actually urgent, tighten the scoring rules. If Warm leads are converting, add a faster callback path. The prompt should get sharper as real lead quality data comes in.