Lead Qualification at the Door: Turning Website Visitors into Booked Calls and Paid Deposits

Twig is an autonomous AI support platform that triages, self-evaluates, and resolves customer support tickets by integrating with tools like Intercom, HubSpot, and Salesforce. For SMBs whose website is the front door — law firms, consultancies, agencies, advisors, specialty practices — Twig is the chat widget that turns the visitor researching at 9pm into a booked consult with deposit by 9:05pm. This post is about the highest-leverage conversion lever on most SMB websites: replacing the contact form with an AI front desk that qualifies and books in real time.

TL;DR: Static contact forms on SMB websites convert at 1–2% — most visitors don't fill them out, and most of those who do never get followed up in time. An AI front desk on the website chat handles the visitor in real time: identifies intent, qualifies the lead, books a call or appointment, captures a deposit, and writes to CRM — all in one conversation. Conversion rates jump to 4–9% of visitors and to 35–55% of conversations engaged. This post is the qualification flow design, the deposit-capture mechanics, the CRM and calendar wiring, and the per-vertical scripts that turn website traffic into booked revenue.

Key takeaways:

• Static contact forms convert at 0.3–1.0% of visitors; AI chat converts at 3–8% of visitors
• Conversation-engaged visitors convert at 35–55% to a booked qualifying action
• Deposit-required flows reduce consult no-shows by 40–60% on high-value bookings
• Qualification questions should be 2–4 max — more reduces completion rate
• Hard vs. soft qualification depends on per-conversion economics; most converge on a middle path
• Twig is the text-side AI front desk for SMB websites; pair with voice AI when callers prefer phone

Why the contact form is a leaky bucket

A typical SMB website's contact form journey:

Step	Conversion
Visitor lands on site	100%
Visitor reads enough to consider contacting	8–18%
Visitor finds and clicks "contact us"	4–8%
Visitor completes the form	1.5–3%
Form submission goes to inbox	1.5–3%
Practice replies within 24 hours	60–80% of submissions
Visitor still interested when reply arrives	50–70%
Conversation continues to booking	40–60%
Visitor → booked consult	0.3–1.0%

The leakiness compounds at every step. Each one loses some fraction of the previous step's output. By the time you net out, less than 1% of visitors become booked.

The structural reasons:

• Contact forms feel like an unknown wait — "will they reply in 5 minutes or 5 days?"
• Friction in filling out forms (name, email, phone, message, captcha)
• Delayed reply means by the time the practice calls, the visitor has moved on
• No real-time interaction means no chance to surface objections or build commitment

The AI chat replacement

The same visitor journey with AI chat:

Step	Conversion
Visitor lands on site	100%
Visitor sees chat widget	100%
Visitor engages with chat	8–15%
Chat conversation completes (5+ turns)	40–60% of engaged
Conversation results in qualifying action (book/deposit/lead)	35–55% of completed
Booked consult actually happens (kept rate, especially with deposit)	80–95%
Visitor → kept booked consult	3–8%

The 5–15× improvement is from three structural changes:

• Real-time response eliminates the "will they reply" doubt
• Conversational qualification surfaces objections in the moment
• In-conversation booking with deposit captures intent before it cools

The qualification flow architecture

A working qualification flow has five stages:

Stage 1: Greet + read intent (10–20 seconds)

The first message in the chat. Either visitor-initiated ("hi, I have a question about...") or AI-initiated proactive prompt after 30 seconds of dwelling.

The AI's job here is to identify intent in 1–2 turns. Possible intents:

• Service inquiry (what do you do, what does it cost)
• Booking intent (I want an appointment / consultation)
• Existing-customer support (I'm a current client)
• Off-topic (job applications, random questions)

Different intents route to different sub-flows. The misrouting failure mode is the AI confidently routing a "tell me about your services" inquiry into a booking flow — annoying the visitor and triggering bounce.

Stage 2: Qualify (2–4 questions, max)

For booking-intent visitors, the AI asks 2–4 questions that filter and prepare:

For a law firm initial consultation:

• "What type of legal matter are you dealing with?" (intent classification + UPL boundary)
• "Has this matter started yet, or are you anticipating something?" (urgency)
• "Anything else you'd like the attorney to know before the consult?" (prep)

For a dental practice new-patient inquiry:

• "Are you looking for a routine cleaning or something specific?" (service classification)
• "Have you had dental care recently, or has it been a while?" (workflow routing — new vs. dormant)
• "Do you have dental insurance you'd like to use?" (insurance verification flag)

For a hotel booking:

• "What dates are you looking at?"
• "How many guests?"
• "Any specific room preferences?"

The qualifying questions serve two purposes: filter unfit prospects, and prepare the human/booked-time to be efficient.

Stage 3: Present the offer + book

Real-time calendar lookup. The AI presents 2–4 available slots:

"Great — based on what you've described, our next available consults with attorney Lin are Tuesday at 2pm, Wednesday at 11am, or Friday at 4pm. Which works?"

Visitor picks. AI confirms the booking, writes to calendar, sends confirmation email/SMS.

Stage 4: Capture deposit (if applicable)

For services where deposit is policy:

"Got Tuesday at 2pm booked — to secure the slot, we collect a $100 refundable deposit that's applied to the consult fee. I'll text you a payment link. The slot's held for 30 minutes while you complete the payment."

The 30-minute hold is a common pattern. Slots that don't get paid within the window release for the next prospect.

For services where deposit is optional, the AI may suggest it as a friction-reducer:

"Would you like to put a deposit down now? It's optional but means you don't have to handle payment at the visit."

Stage 5: Confirm + cross-sell

Booking confirmed, deposit captured (or noted). AI:

• Sends confirmation
• Provides prep details ("bring X, Y, Z")
• Optionally suggests an add-on (upsell playbook)
• Writes the full lead record to CRM with intent, qualifying answers, transcript

Hard vs. soft qualification — when each makes sense

Two patterns:

Hard qualification

3–4 questions, deposit required, structured intake. Conversion of engaged-chat to booking: 25–40%. Conversion of booked consults to kept consults: 90%+.

Best for:

• High-value services ($500+ consult)
• High no-show prone industries (legal initial consultations, financial advisory)
• Practices with limited slot supply that need to filter
• Specialty services with significant prep requirements

Soft qualification

1–2 questions, no deposit, easy booking. Conversion of engaged-chat to booking: 45–65%. Conversion of booked consults to kept consults: 65–80%.

Best for:

• Low-value or first-visit-free consults
• Practices with abundant slot supply (newer practices building volume)
• High-volume verticals where filtering is less important
• Industries where deposit-asking would create friction

The middle path (most SMBs end up here)

2–4 questions, deposit required for high-value but not low-value, intelligent default.

The right pattern is determined by per-conversion economics. If a consult yields $4,000 in expected value, soft qualification capturing more volume is worth the lower kept rate. If a consult yields $400 and is 70 minutes of provider time, hard qualification protects the practice's most expensive resource.

The deposit-capture mechanics

Common payment processors that integrate cleanly with AI front desks:

Processor	Strengths	SMB pricing
Stripe	Best developer experience, Payment Links	2.9% + $0.30
Square	Best for retail-adjacent SMBs	2.6% + $0.10
AuthorizeNet	Older, broad acceptance	Tiered, ~2.9%
PayPal	Trust signal for consumers	2.9% + $0.30

For deposit-capture flows:

• Use Payment Links (Stripe, Square Invoices) — no need to host payment forms
• Set explicit refund policy (refundable until 24h before appointment is common)
• Webhook to AI on payment success → confirms booking; on timeout → releases slot
• Tokenize for future bookings if customer opts in

The CRM write-back

Every qualified conversation should write to CRM:

• Contact record (name, email, phone, source = AI chat)
• Lead record with intent classification
• Conversation transcript (PII-screened)
• Qualifying answers structured
• Booked appointment (linked to calendar)
• Deposit status
• Next action (follow-up date, escalation flag)

For SMBs with light CRM, HubSpot free tier handles this; for more sophisticated stacks, Salesforce or specialty PMSs (Clio for law firms, Dentrix for dental, etc.) integrate via API or webhook.

Twig's text-side architecture for autonomous resolution uses the same write-back pattern — every conversation produces a structured record in CRM, not just a free-text note.

Per-vertical scripts

Law firm initial consult

"Hi — I help with initial consult inquiries. Were you looking to schedule a consult, or did you have a quick question I can help with?"

"(book) Got it. To get you to the right attorney, can you tell me briefly what type of matter? Family law, criminal defense, employment, real estate, business, or something else?"

"(employment) Got it. Has this matter started — for example, have you been terminated or received a notice — or is this something you're anticipating?"

"(terminated last week) Okay — let me get you booked with attorney Patel who handles employment matters. I have Tuesday at 2pm, Wednesday at 11am, or Friday at 4pm. Which works best?"

"(Tuesday) Great, Tuesday June 4th at 2pm with Attorney Patel. We do a $200 refundable consult fee — I'll text you a payment link. Once paid, you'll get a confirmation and intake form. Anything else?"

B2B SaaS demo request

"Hi! Looking around for a CRM or a specific question?"

"(CRM evaluation) Got it. Quick question to send the right info — are you replacing an existing system or starting fresh?"

"(replacing Pipedrive) Makes sense — happy to walk you through how we handle migrations. Want to grab 20 min with our team this week? I see openings Wed at 1pm Pacific or Thursday at 10am."

Dental practice new patient

"Welcome! Are you a current patient or new to the practice?"

"(new) Great. Are you looking to schedule a cleaning, or do you have something specific going on — like a tooth that's bothering you?"

"(specific tooth pain) Got it — let me see if we can get you in soon. Where exactly is the pain, and how long has it been bothering you?"

"(upper right, 2 days) Okay. Dr. Chen has an emergency slot tomorrow at 10am or 3pm. Which works?"

Hotel reservation

"Hi — looking for a stay date?"

"(July 12–14, 2 guests) Got it. We have a couple of options for those nights — standard king at $185/night or deluxe with balcony at $235/night. Either preference?"

Where Twig fits

For SMBs, Twig is the text-side AI front desk on the website chat — handling exactly the qualification + booking + deposit flow described above. The wiring:

• Twig's website chat widget runs the qualification conversation
• Twig's calendar integration (Google, Microsoft, Calendly, Acuity, or PMS-specific) provides real-time availability
• Twig's payment processor integration generates and tracks deposit links
• Twig's CRM write-back (HubSpot, Salesforce, Intercom) updates lead and contact records
• Twig's self-evaluation layer prevents over-promising on services not offered or misclassifying intent

Pair with a voice AI vendor for callers who arrive at the website but prefer phone. The voice + chat front desk pair under shared knowledge and calendar.

The bottom line

For most service-based SMBs, the contact form is the single biggest leak in the website-to-revenue funnel — losing 95–99% of intent-bearing visitors to friction and delayed response. An AI front desk on website chat recaptures the bulk of that loss, converting 3–8% of visitors instead of 0.3–1%, at no incremental cost per visitor.

The deployment work is modest: chat widget on the site, calendar and payment integration, 2–4 qualifying questions designed per vertical, and CRM write-back wired. Most SMBs see the new lead flow within 24 hours of going live, and the booked-with-deposit consult rate stabilizes within 30 days. From there it's a steady-state revenue lever, not a project.