🎯 Lead Qualification BANT+ Script
Purpose
Produce a structured, conversational qualification script (usable by a live agent, ISA, voicemail follow-up, chatbot, or email sequence) that reliably separates serious buyers and sellers from browsers — and outputs a 0–100 lead score with a routing recommendation — using an extended BANT framework adapted for real estate.
When to Use
Use this skill when a new lead comes in from any channel (portal inquiry, open-house sign-in, website form, referral, social DM, past-client reactivation), when setting up an AI chatbot or SMS responder and needing a qualification flow, when coaching a new ISA or assistant on what to ask, or when reviewing a long-dormant CRM list and deciding where to reinvest follow-up time. Pair with buyer-follow-up-sequence.md — this skill decides whether a lead enters the sequence; that skill runs the sequence.
Required Input
Provide the following:
- Lead source — How the lead arrived (Zillow, open house, referral, IG DM, past client, cold call, etc.)
- Available context — Any data already known (property they inquired about, timeline phrases they used, geography, name, phone, email)
- Conversation channel — How the qualification will happen (live phone, SMS, website chatbot, email, in-person open house, voicemail drop)
- Buyer or seller — Which side the lead is on (or "unknown" if the lead could be either)
- Agent market — Primary geography and typical price band so the script can calibrate "serious" thresholds appropriately
- Handoff policy — What the agent wants to happen with hot leads (immediate call within 5 minutes, same-day text, book showing, send property match)
- Compliance constraints — TCPA / DNC / state-specific consent rules that apply, and whether the conversation is recorded
- Recon depth available — How much time exists before the first touch must go out, and which property/market data sources the agent can actually reach (MLS, tax roll, CRM history, a cached ZIP market profile). Defaults to Fast Pass if unspecified.
Instructions
You are a real estate sales enablement AI assistant. Your job is to design a qualification conversation — not an interrogation — that earns the lead's trust while extracting the signals needed to score and route them correctly.
Before you start:
- Load
config.ymlfrom the repo root for brand voice, agent signature, and routing rules - Reference
knowledge-base/regulations/for TCPA, DNC, and state-specific disclosure requirements (recording consent, AI disclosure) - Reference
knowledge-base/best-practices/for channel-specific norms (SMS character limits, chatbot message length, voicemail max duration)
Process:
-
Run a recon pass before the first message — at a depth the clock allows. Speed-to-lead is the single highest-leverage variable in this workflow: a two-minute reply routinely outperforms a two-hour reply, and the agents who convert are the ones whose first touch already sounds informed about the property. But research depth and response speed trade against each other, so declare which pass you are running and never let the deep pass delay the first touch.
Fast Pass (default) Deep Pass Budget ≤ 2 minutes 5–15 minutes When Any inbound lead — always run this first High-value lead (top of the agent's price band), seller/listing lead, referral, or any lead scoring 80+ on the Fast Pass Sources Whatever is already at hand: the inquiry itself, the listing record, the agent's CRM history, a cached ZIP market profile Adds verified property-level records — MLS/tax roll/county assessor, public price history, prior listing photos and days-on-market Pulls Property basics (beds/baths/sqft/year built), neighborhood context, current ZIP market snapshot, any prior contact with this person Everything in Fast Pass plus verified tax/ownership data, full price and listing history, length of ownership, equity posture, condition signals from prior photos Feeds The warm opener + a first-cut lead score A pre-appointment brief, and hands directly to cma-presentation-generator.md(seller) orbuyer-follow-up-sequence.md(buyer)Rules for the recon pass:
- Never block the first touch on the Deep Pass. Send the Fast-Pass-informed opener, then go deep while you wait for the reply.
- Label every fact by confidence. Anything from the Fast Pass is a working assumption; anything from the Deep Pass is verified. Never state an unverified figure to the lead as fact — "looks like about 1,900 square feet, is that right?" invites a correction and builds trust; asserting it and being wrong destroys both.
- Cache market research by ZIP, not by lead. Neighborhood trends, school context, inventory, days-on-market, and median price move slowly and are shared across every lead in that area. Write them to a reusable ZIP-level market profile and refresh on a cadence (monthly is usually enough), rather than re-researching the same neighborhood for each new lead. The library gets faster the more leads the agent works in their farm.
- Recon informs the opener; it does not script the conversation. The point is to sound like someone who already knows the block — not to recite a dossier at a stranger, which is unsettling and reads as surveillance.
- Fair housing applies to research, too. Never pull, infer, or record protected-class signals about the lead or the neighborhood (demographic composition, "family-friendliness," "good schools" as a proxy). Neighborhood context means inventory, price trends, and days-on-market — not who lives there.
-
Open with context, not a form — The first message must reference how the lead arrived and what they showed interest in, and should carry one specific, verifiable detail from the recon pass. Cold openers ("Hi, are you still looking to buy a home?") fail. Warm openers ("Saw you asked about the place on Elm — want me to send you a couple more like it?") get engagement. Recon-informed openers ("Saw you asked about the Elm St place — it's the 1926 Craftsman that just came back on after a price cut. Want two or three more like it?") get replies.
If using AI in any automated channel, disclose it clearly and early per local rules (example: "Heads up — I'm [Agent]'s assistant and part of this conversation uses AI. I'll loop [Agent] in directly for the important parts.").
-
Work through BANT+ in natural conversational order — Don't ask all four at once. Sequence by what's easy to answer first:
- N — Need (easiest to answer, establishes the "what"):
- Buyers: property type, bed/bath, must-haves, desired area(s), current living situation
- Sellers: property address/area, reason for selling, target net, condition
- T — Timeline (reveals urgency):
- Buyers: "When would you want to be in the new place?" (0–30 days = hot, 30–90 days = warm, 3–6 months = nurture, 6+ months = long-term)
- Sellers: "What's driving your timeline — is there a specific date you need to be out / sold by?"
- B — Budget (or, for sellers, expected net):
- Buyers: "Have you talked to a lender yet?" → pre-approval amount → down payment comfort zone. Never ask "what's your budget" as the first budget question; ask about lender first.
- Sellers: "Do you have a number in mind for what you'd need to walk away with?" (not "what do you think it's worth") — opens a different conversation than pricing debate
- A — Authority (decision-making structure):
- "Who else is in on this decision — spouse, partner, family?" / "Are you working with another agent currently?"
- F — Fit (extended) — Does the agent actually want this lead?
- Geography in service area, price point in agent's sweet spot, realistic for market conditions
- I — Intent (extended) — Behavioral signals:
- Already pre-approved? Already toured homes? Already signed a buyer rep agreement? Already had a listing appointment? Already talked to another agent?
- N — Need (easiest to answer, establishes the "what"):
-
Adapt by channel — Rewrite the same qualification flow for the channel the agent specified:
- Live phone (5–7 min) — Conversational, open questions, note taking
- SMS (3–5 messages total) — Short, one question per message, casual tone
- Website chatbot (8–12 turns max) — Button options plus one free-text field per turn
- Email (1 initial message + branching follow-ups) — Fewer but deeper questions, designed to get a reply
- Voicemail drop (30–45 seconds) — Skip qualification; goal is callback. Open with referral/context and end with a specific reason to call back.
- In-person (open house, 2–3 min) — Sign-in card covers N+T; agent covers B+A+F in person
-
Score the lead — After the qualification exchange, score on a 0–100 scale:
Dimension Weight Hot signal Warm signal Cold signal Budget / Pre-approval 25 Pre-approved, lender confirmed Talking to lender No lender yet Authority 10 Decision-maker, no other agent Co-decision, no other agent Already with another agent Need 15 Clear criteria, specific area General area, flexible Vague, "just looking" Timeline 25 0–60 days 2–6 months 6+ months or "whenever" Fit 10 In geography + price band Adjacent geography/price Outside service area Intent 15 Toured homes, ready to act Researching actively Browsing portals only Routing rule:
- 80+ → Hot. Alert agent for contact within 5 minutes. Offer showing/appointment.
- 50–79 → Warm. Enter long-term nurture via
buyer-follow-up-sequence.md. - <50 → Cold. Drop to low-touch drip (monthly market updates only).
-
Write the outputs — Deliver five things:
- The recon brief — A short block from Step 0: what is known about the property and ZIP, each line marked verified or assumed, and the one detail worth putting in the opener
- The script itself — Ready to paste into the chosen channel, with branching (what to say if they answer yes/no/skip)
- A one-page cheat sheet — For a live agent/ISA, a single-page reference showing what to listen for
- The scoring rubric filled in with blanks — So the agent or automation can score the lead after the call
- A sample handoff message — What the agent should say first when taking over a hot lead from qualification
-
Flag compliance and etiquette — Call out:
- TCPA / consent language needed for the channel
- AI disclosure language required by state/brokerage
- Any question that crosses into fair housing protected-class territory (never ask about family size, religion, national origin, disability status, marital status)
- Any question that's premature for the channel (e.g., don't ask for SSN or pre-approval docs in an SMS qualification)
Critical rules:
- Never ask fair-housing protected-class questions (family composition, kids, schools for kids, religion, national origin, marital status, disability) during qualification. If the lead volunteers them, don't store as structured data.
- Never quote what their house is worth or what they'll qualify for during qualification. Defer pricing to a CMA appointment and financing to the lender.
- Never promise response times the agent can't honor.
- Always disclose AI where required by law or brokerage policy.
- Never push past three unanswered messages on SMS before a cooldown.
- Never let recon delay the first touch. A Fast-Pass reply in two minutes beats a Deep-Pass reply in two hours, every time.
- Never state an unverified recon fact to the lead as though it were confirmed. Ask; don't assert.
- Never research or record protected-class characteristics of a lead or a neighborhood, at any recon depth.
Example Output
Input summary: Lead source: Zillow inquiry on 4829 Glenalbyn Dr ($875K). Channel: SMS. Side: Buyer. Market: Highland Park LA, $600K–$1.4M band. Handoff: agent call within 5 min on hot leads. Recon depth: Fast Pass (lead came in 4 minutes ago; MLS + cached 90042 market profile on hand).
Output (abbreviated):
Recon brief — Fast Pass, 90 seconds:
- 4829 Glenalbyn Dr — 3BR/2BA, ~1,640 sqft, 1926 Craftsman (verified — own listing)
- 22 days on market; one price cut, $899K → $875K (verified — own listing)
- 90042 snapshot: median $842K, 31 DOM, inventory flat MoM (cached ZIP profile, refreshed 7/1 — assumed current)
- No prior contact with this lead in CRM (verified)
- Opener hook: it's the Craftsman, and it just took a price cut — lead a comparable set with that framing.
- Deep Pass queued: if the lead scores 80+, pull tax roll + full price history before the callback.
SMS Script — 5 messages max:
Msg 1 (immediate): "Hey — Jamie Chen here, I'm the listing agent on 4829 Glenalbyn, the '26 Craftsman. Noticed you inquired on Zillow. Want me to send you two or three similar Highland Park homes in that price band?"
Msg 2 (if "yes"): "Great — quick so I can match what fits: are you hoping to be in the new place in the next couple months, or more of a this-summer thing?"
Msg 3 (after timeline): "And have you had a chance to chat with a lender yet, or still early days?"
Msg 4 (after budget signal): "Last one — are you working with another agent already, or open to having me run point on this?"
Msg 5 (close): "I'll send over a few Highland Park options tonight. Want to book 15 min tomorrow to walk through what you're looking for?"
Cheat sheet (for agent reference):
- Listen for: "pre-approved," "we're moving in [month]," "looking in [specific area]," "not working with anyone yet"
- Warning signs: "just browsing," "sometime next year," "our agent is showing us this Saturday"
Scoring example (filled):
- Budget: pre-approved at $950K → 25
- Authority: couple, no other agent → 10
- Need: specific to Highland Park, 3BR → 15
- Timeline: 60 days (hot band is 0–60) → 25
- Fit: in market + price → 10
- Intent: toured 2 homes already → 15
- Total: 100 → HOT → alert Jamie
Handoff message: "Hey [Lead] — Jamie here, taking over from the quick intro earlier. Have a few Highland Park options teed up for you. What's a good time tonight or tomorrow morning for a 10-min call?"
Compliance flags:
- Opening SMS includes agent identity ✓
- No AI disclosure needed (agent is sending personally)
- No fair housing protected-class questions asked ✓
- TCPA: Zillow inquiry constitutes prior express consent for related SMS — log consent source in CRM