Conversational AI Scripts for Lead Response: Templates by Industry

A conversational AI script is the set of messages, questions and branching rules that tells an AI how to greet a new lead, qualify it, book the next step and hand off to you. The best ones read like a short, helpful text from a person who knows your business. Below you will find scripts you can adapt for real estate, contracting, insurance and financial advising, plus the rules for unclear answers, objections and US consent.

Astreauxcalendly.comRespond to Leads in Your VoiceAstreaux uses conversational AI to learn your business’s voice and respond to new leads instantly, helping service professionals manage follow-up.Book a call

TL;DR

• A good lead-response script has six parts: greeting, one qualifying question at a time, a tailored answer, a booking offer, a recovery path and an honest handoff to a person.

• Keep each message short, use the lead’s own words back to them, and ask for one thing per message.

• Write industry scripts around the decision the lead is making: viewing a home, scheduling an estimate, getting a quote, or starting a planning conversation.

• In the US, check consent before AI-voiced calls or marketing texts. The FCC treats AI-generated human voices as artificial voices under the TCPA.

• Test the whole conversation, including the awkward replies, before you switch it on.

What a conversational AI script actually is

A traditional sales script is a monologue you read. A conversational AI script is closer to a map. It holds your approved wording, the questions you need answered, the facts the AI may state, and the points where it must stop and bring in a human.

That matters for service professionals because most leads arrive when you are busy: showing a house, standing on a roof, sitting with a client. A script lets the first reply go out in seconds, in your voice, while the lead is still looking at their phone.

A script has five building blocks:

1. Opening. Who is writing, why, and what the lead can expect.

2. Qualifying questions. The few facts you need to decide what happens next.

3. Answers from approved sources. Service areas, hours, process, what an appointment involves.

4. Next step. A booking link, two time options or a callback request.

5. Exit ramps. What to do when the AI does not understand, when the lead objects, and when they want a person.

If you want the bigger picture of how this fits a revenue process, our guide to conversational AI for sales walks through it from the leadership side.

The six-part structure every lead-response script needs

1. A greeting that references the inquiry

Name the source and the request. A lead who filled out a form about a three-bedroom home should not receive a generic welcome.

Hi {{firstName}}, this is {{agentName}}'s assistant at {{businessName}}. Thanks for your question about {{propertyOrService}}. I can help get you answers right away. Is now a good time to chat by text?

Say plainly that it is an automated assistant working for you. It sets expectations and keeps trust intact when you step in later.

2. One qualifying question at a time

A five-question block reads like a form, and leads abandon forms. Ask one question, respond to the answer, then ask the next. Choose questions that change what you do next:

• Timeline: “When are you hoping to get started?”

• Location: “Which area or ZIP code is this for?”

• Scope or need: “What are you trying to get done?”

• Decision-makers: “Will anyone else be part of the decision?”

3. A tailored answer

Once the lead has told you something, say it back and connect it to what you do. This is the step that separates a conversation from an autoresponder. The principles behind it overlap with how to humanize AI-written text with clear instructions: specific details, plain wording and a natural rhythm.

Got it, you’re looking at a kitchen remodel this spring and the cabinets are the main piece. That’s something we handle often. The next step is a short on-site visit so we can measure and give you a real number.

4. A booking offer with low friction

Offer two concrete times or a link, not an open-ended “let me know.”

Would Tuesday at 4:00 PM or Thursday at 10:00 AM work better for the visit?

5. A recovery path

When the lead says something the script does not cover, the AI should say so and narrow the question instead of guessing.

I want to get that right. Are you asking about pricing, timing, or something else?

6. An honest handoff

When the lead asks for a person, or the question needs your judgment, the AI should confirm and hand off.

Absolutely. I’ll let {{agentName}} know right now and they’ll reach out today. What’s the best number and time?

Real estate scripts

Real estate leads come from portals, ads, open houses and your website. They are often browsing several agents at once, so speed and relevance decide who gets the reply.

New inquiry on a listing

Hi {{firstName}}, thanks for asking about {{address}}. I’m {{agentName}}'s assistant. Are you hoping to see it in person, or do you have a question about the home first?

If they want a showing:

Great. Are you working with another agent right now, or would you like {{agentName}} to help? Either way, I can find a time that works for a tour.

The agency question is worth asking early, but phrase it as an offer to help, not an interrogation.

Qualifying a buyer

Use short questions in this order:

1. “What price range are you comfortable with?”

2. “Do you have a pre-approval, or would an introduction to a lender be useful?”

3. “What’s your ideal move-in timeframe?”

4. “Which neighborhoods are at the top of your list?”

Each answer should earn a short acknowledgment before the next question. Skipping the acknowledgment is what makes a script feel robotic.

Seller inquiry

Thanks for reaching out, {{firstName}}. Are you thinking about selling in the next few months, or just curious what your home might be worth right now?

A seller who is “just curious” belongs in a longer nurture track. A seller with a timeline belongs on your calendar. The script should route them differently.

Open house follow-up

Hi {{firstName}}, it was great meeting you at {{address}} today. What did you think of the home? If you’d like to see similar places nearby, tell me what mattered most and I’ll pull a few.

Contractor scripts

Homeowners requesting estimates want to know three things: do you do this work, are you available, and what happens next. A contractor script should answer those in under five messages and leave you with a qualified appointment.

Estimate request

Hi {{firstName}}, thanks for contacting {{businessName}} about {{projectType}}. I have a few quick questions so we can send the right person. Is the property in {{serviceArea}}?

Then collect, one at a time:

1. Project type and rough scope (“Replacing a few windows or the whole house?”)

2. Property address or ZIP code

3. Timeline (“Is this urgent or are you planning ahead?”)

4. Photos, if useful (“If you can text a photo of the area, that helps us prepare.”)

Scheduling the estimate

Thanks, that’s everything I need. Our estimator can come out Wednesday between 9 and 11 AM or Friday between 1 and 3 PM. Which window is better?

After booking, send a confirmation and a reminder. A clear confirmation lowers no-show risk:

You’re set for Wednesday, 9 to 11 AM. Please make sure someone is home and the work area is accessible. Reply here if you need to change the time.

Out-of-area or out-of-scope leads

A script should decline politely without leaving the lead stranded.

Thanks for thinking of us. That project is outside what we currently handle, so I don’t want to waste your time. Good luck with it.

Only include referral language if you actually have referral relationships to offer.

Insurance agent scripts

Quote requests are time-sensitive. The lead is comparing, and the first agent to respond with something useful often gets the conversation. Our guide to the insurance lead follow-up cadence covers timing in detail; here is the wording.

Instant quote-request reply

Hi {{firstName}}, thanks for requesting a quote for {{coverageType}}. I’m {{agentName}}'s assistant. To get you accurate options, can I confirm a couple of details? First, what’s the ZIP code where you need coverage?

Collecting the basics

Ask only what you need to schedule a conversation, not everything a quote requires. Sensitive personal details are better collected by you on a call or through a secure form than by text.

• Type of coverage

• Renewal date of any current policy

• Best time for a call

Follow-up when the lead goes quiet

Hi {{firstName}}, just checking in on your {{coverageType}} quote. Do you have any questions I can answer, or would a quick call this week be easier?

Keep the language informative, not pressuring, and avoid promising specific rates or coverage by text. Those statements belong to a licensed agent.

Financial advisor scripts

Advisory leads move slowly and decide on trust. The script’s job is to start a conversation about the prospect’s situation, then keep helpful contact going until they are ready.

First response

Hi {{firstName}}, thanks for reaching out to {{firmName}}. {{advisorName}} would like to learn what prompted your inquiry. What’s the main thing you’d like help with right now?

Discovery-style questions

Focus on the person, not a product pitch:

1. What is your situation today?

2. What goal are you working toward?

3. What have you tried so far?

4. What would a good outcome look like?

These map to the flow of a real discovery call, so the AI is collecting context your advisor can use rather than screening people out.

Nurture after a call

For prospects who did not move forward, a spaced sequence works better than one big follow-up. A sample rhythm we suggest testing: a thank-you the day after, a short useful resource a few days later, a check-in after a week, then lighter touches over the following weeks. Adjust it to your clients, and have your compliance team approve the wording, since advisory communications are regulated.

Hi {{firstName}}, I was thinking about what you said regarding {{topic}}. Here’s a short article that speaks to it. If it would help, {{advisorName}} is happy to pick the conversation back up.

Scripts for the moments that break most chatbots

The opening of a script is easy. The difficult part is everything unscripted.

Unclear or off-topic replies

When a lead replies “maybe” or sends something unrelated, restate the last question in a simpler form and offer choices.

Sorry, I want to make sure I understand. Are you looking to schedule a visit, or do you have a question first?

After two failed attempts, stop looping and hand off.

Common objections

Rescheduling and cancellations

No problem, things come up. I can move your visit to {{option1}} or {{option2}}. Which works?

Make changing a booking as easy as making one. Leads who find rescheduling difficult tend to cancel silently.

“Can I talk to a person?”

This request should always work, instantly, with no argument. Confirm, collect a callback number if needed, and notify you through your usual channel.

Of course. I’ll let {{agentName}} know right away. What’s the best number to reach you?

How to keep scripts from sounding robotic

• Use contractions and plain words. “We’ll” beats “we will.”

• Vary your openings. If every message starts with “Thanks for reaching out,” it reads like a template.

• Reflect the lead’s words. If they say “fixer-upper,” you say “fixer-upper.”

• Keep messages short. One or two sentences, one question.

• Write like you talk. Read each line aloud. If you would not say it to someone at an open house, rewrite it.

• Train on your real replies. Feed the AI your best past messages so it matches your voice. Our AI suggested replies guide explains how to review drafts before they go out.

Privacy, consent and follow-up rules in the US

Scripts live inside legal limits. Before you automate outreach, confirm your consent practices.

• AI-voiced calls. The FCC ruled on February 8, 2024 that calls using AI-generated human voices count as artificial voice calls under the TCPA, so they require the called party’s prior express consent.

• Texts. Marketing texts also fall under TCPA consent rules. Capture consent clearly on your forms, include opt-out language, and honor “STOP” immediately. Our TCPA compliance guide for SMS goes through the steps.

• Sensitive data. Do not ask leads to send financial account numbers, health details or similar information by text.

• State rules. Some states add their own calling and texting rules, and insurance and advisory work carries industry regulation. Confirm with your attorney or compliance officer.

For how inbound automation and compliance fit together, see our walkthrough on inbound lead automation.

Our perspective

Most lost leads are not lost to a competitor. They are lost to silence. A lead who asks a question at 7 PM and hears back the next afternoon has often already talked to someone else.

That is why we built Astreaux around instant, conversational replies that learn the voice of each business. A script is where your expertise goes in: the questions you ask, the way you explain things, the point where you want to take over. The AI’s job is to carry that wording to every new lead the moment they arrive, then book the appointment and keep nurturing.

We design Astreaux for real estate agents, contractors, insurance agents, financial advisors, therapists and mortgage brokers, and it connects with more than 7,000 apps, so your scripts can work alongside the CRM and ad platforms you already use. Start with one script, one lead source and one clear goal. Then expand.

Jamaal

Put your scripts to work with Astreaux

If you would rather not build every message by hand, Astreaux AI is built for service professionals who want instant lead replies, appointment booking, nurturing and bulk SMS campaigns in one place. You train it on your voice, connect your CRM and ad platforms, and it begins conversations the moment a lead comes in.

See how it works on the Astreaux AI homepage, or read our guide to running a conversational AI pilot for leads to plan your first few weeks.

Test before you launch

A script that reads well on paper can fail in a live conversation. Before launch:

1. Run the happy path. Play a lead who answers every question cleanly and books.

2. Run the messy paths. Reply with one-word answers, typos, off-topic questions and a request for a person.

3. Check every handoff. Confirm that you receive the alert, with the conversation history.

4. Review the first 20 real conversations. Look for places where leads stopped replying, and rewrite the message just before.

5. Track first-touch time and booking rate. Those two numbers tell you quickly whether the script is helping. Our guide to AI lead qualification covers what to measure.

Change one thing at a time so you know what caused the difference.

FAQ

How long should a conversational AI script be?

Keep each message to one or two sentences and one question. The whole script can have many branches, but any single conversation should reach a booking or a handoff in about five to eight exchanges. If a lead needs more, hand off to you.

Can I use the same script for every lead source?

Use the same structure but change the opening. A lead from an ad, a listing portal and a referral each expect different context, so reference the source in your first message. Your qualifying questions can stay mostly the same.

Should the AI say it is not a person?

Yes. Introducing the assistant as automated sets honest expectations and avoids surprise later. For calls with AI-generated voices, US consent rules also apply, so check them before dialing.

What should the AI never say?

It should not promise prices, rates, coverage, returns or outcomes that only you or a licensed professional can give. Write those as “{{agentName}} can walk you through the details” and move to booking a conversation.

How do I get the AI to match my voice?

Gather several of your best real replies, note the phrases you use often, and give the AI those as examples. Review its drafts for the first few weeks and edit anything that does not sound like you.

Sources

• FCC: TCPA applies to AI technologies that generate human voices