Most chatbot content treats the bot as a cost saver that deflects tickets and shrinks the queue. This playbook covers the other use. A chatbot can be a revenue channel that answers visitors while their interest is live, qualifies them in conversation, and books the meeting before they leave the tab.
Conversational marketing is the practice of converting website visitors into qualified pipeline through real-time conversation on the page, instead of static forms and delayed follow-up. Drift popularized the term and built a category on it. Salesloft acquired Drift in February 2024, and the market has since moved from live SDRs toward AI chatbots trained on the company's own content. Options for teams migrating off Drift-era tooling are compared in Drift alternatives.
iShort answer
A sales and marketing chatbot turns live interest into pipeline in one session: Engage with a useful answer, Qualify through conversation, Book the meeting before the visitor leaves. Ask only the three to five questions sales cannot work without, one at a time, and track capture rate, qualification rate, and meetings booked in that order. The bot covers the top of the funnel around the clock, and a human takes over fast once intent is clear.
Engage, Qualify, Book: the three jobs a revenue chatbot does in one session, in order. Each stage has one metric, so a sagging number points at the stage that owns it.
Engage
Open with usefulness, not a pitch: answer the question the visitor actually has, from your own content, in seconds. Metric: share of conversations that yield contact details (capture rate).
Qualify
Ask the three to five questions sales cannot work without, one at a time, each after the bot has given something first. Metric: share of captured leads that meet the bar (qualification rate).
Book
Hand qualified, high-intent visitors a calendar in the same session, and route hot ones to a human on demand with full context. Metric: meetings booked.
The case for speed is old and well measured. A Harvard Business Review study of online sales leads found that firms contacting a lead within an hour were nearly seven times as likely to qualify it. The comparison group waited just one hour longer. The same audit of 2,241 US companies found that 23% never responded to a web lead at all. A chatbot closes that gap by answering in seconds.
Why do forms lose leads that conversations keep?
A form bets that the visitor's interest will survive the wait. The visitor has a question, and the form offers an email field and a promise of a reply. Interest decays every hour until that reply arrives, and high-intent visitors are usually comparing options right now.
A conversation reverses the order. The bot answers the question first, from your own content, while the visitor is still on the page. The request for contact details comes after the visitor has already received something useful.
Three advantages follow:
- Timing. A conversation happens at the moment of intent, including 2 a.m. in a time zone where nobody on the team is awake. With a platform that answers in 95+ languages, business hours stop limiting pipeline.
- Order. A form collects data before it gives anything back. A conversation gives value first and asks second, and capture rates differ on this alone.
- Detail. A form captures fields. A conversation captures the visitor's own words: the problem, what they tried, and which vendors they are comparing. A seller would otherwise spend the first call collecting that.
What is the E.Q.B. funnel?
The E.Q.B. funnel is a three-stage sequence: Engage, Qualify, Book. It compresses the classic marketing funnel into one session. Engagement and education become one stage, because for a chatbot both mean answering questions well.
Each stage has one job and one metric. The sections below cover each stage in order, and the metrics get their own section because that is where most teams' tracking breaks down.
How should a sales chatbot open the conversation?
It should open by answering the visitor's question. A widget that pops up with "Want to book a demo?" behaves like a form with animation, and visitors treat it as one. The quality of the first answer decides everything after it.
That makes the bot's training content a revenue decision. A bot trained only on the marketing site answers marketing questions.
Buying questions get answered in the docs, pricing details, comparison pages, and video walkthroughs. SiteGPT trains on 12 or more content source types, including help centers, documentation, files, cloud drives, and YouTube. The steps for loading and testing that content are in the guide to creating an AI chatbot, and the lead-capture build is in the lead generation chatbot tutorial.
Two engagement rules matter most:
- Answer before asking. The bot's first turns give answers, links, and specifics. Requests for contact details come after the conversation has shown it is worth continuing.
- Match the page's intent. A visitor on the pricing page already knows what the product does. Welcome messages and prompts should assume the intent the page implies.
What qualification questions should a chatbot ask?
Ask only the three to five answers sales cannot work without: the problem, company size or volume, timeline, and current tooling. Everything else turns qualification into an interrogation.
Four rules keep qualification conversational:
- One question per turn. Three questions in one message is a form again. Alternate between giving an answer and asking a question.
- Sequence by commitment. Open with the question the visitor wants to answer anyway (what they are trying to solve). Ask for contact details last, once the exchange has value behind it.
- Skip what you can infer. Company size and industry often come from the email domain. Asking anyway signals a script.
- Let answers branch. Tell an out-of-scope visitor so, early and politely. A polite no costs one lead. A mis-sold demo costs a seller an hour and the visitor's trust.
Ready-to-paste qualification scripts, including the question order, are in the chatbot templates library. Scoring thresholds and routing rules are covered in the lead qualification chatbot guide.
How do you book the meeting in the same session?
Put the calendar inside the conversation. Qualified pipeline often dies between "yes, let's talk" and a confirmed slot. Each follow-up email, scheduling thread, and time zone conversion gives interest another chance to fade.
Match the booking ask to what the conversation has already established:
- Qualified and high intent: offer the calendar directly.
- Qualified but early: offer a lower-commitment close, such as a resource, a follow-up, or permission to email.
- Not a fit: give an honest redirect.
The full setup, including calendar integration, is in the appointment booking chatbot tutorial.
Which metrics show whether a sales chatbot works?
Track three ratios in sequence, each owned by one funnel stage. The first ratio that sags points at the stage to fix.
| Metric | Definition | When it sags, look at |
|---|---|---|
| Capture rate | Contact details collected ÷ meaningful conversations | The Engage stage: the ask comes too early, or the answers are not worth an email |
| Qualification rate | Leads meeting the bar ÷ leads captured | The Qualify stage, or upstream traffic: wrong visitors, or questions that fail to filter |
| Meetings booked | Slots confirmed ÷ qualified leads | The Book stage: friction between yes and the calendar, or an ask that does not match intent |
Define a "meaningful conversation" strictly, as several turns with a real question. That keeps drive-by widget opens out of the denominator. Baseline every ratio against the form funnel you are replacing.
Read transcripts as well as ratios. The conversations where visitors leave mid-qualification show which question causes the friction, and no dashboard reports that. Vendor case studies quote capture and conversion benchmarks freely. Treat those as marketing until you reproduce them on your own traffic.
When should the chatbot hand off to a human?
Hand off fast when the visitor asks for a person, when deal signals exceed the bot's brief (custom pricing, procurement, security review), or when the visitor is frustrated. The chatbot owns the top of the funnel. Selling belongs to people, and the boundary should be set on purpose.
On SiteGPT, the chat escalates when the visitor asks for a person, and the team is notified. On Growth and above it can also route into chat tools such as Slack and Crisp, with Zendesk available on Scale and above. A good handoff is immediate and carries the conversation with it, so the buyer never repeats what they told the bot. The human handoff comparison evaluates platforms on exactly this.
The same boundary protects marketing claims. A bot trained on your own content answers from what is true today. It should not be scripted into promises, discounts, or roadmap commitments that a human would not make in writing. The ways a bot drifts into invented commitments, and the guardrails that stop it, are covered in chatbot hallucinations.
How do you build a sales and marketing chatbot?
Build it in this order:
- Train the bot on everything pre-sale. Site, docs, pricing pages, help center, and video. The breadth of this content caps how well the Engage stage performs.
- Design the qualification set. Three to five questions, each passing the canceled-meeting test, asked one per turn.
- Wire the booking path. A calendar in-session for qualified visitors, softer closes for early ones, and honest redirects for poor fits.
- Set the handoff rules. On-demand escalation, deal-signal triggers, and notifications into the tools sellers already watch.
- Instrument the three ratios. Baseline them against the form funnel and read transcripts weekly.
Each layer has its own tutorial: lead generation, qualification, and booking.
Frequently asked questions
What is conversational marketing?
It is the practice of converting website visitors into qualified pipeline through real-time conversation on the page, instead of static forms and delayed follow-up. Drift popularized the term. The practice outlived the vendor because interest still decays while forms wait.
How does a chatbot generate leads?
It answers the visitor's blocking question immediately, then asks for contact details once the conversation has shown value. Qualification happens along the way, and high-intent visitors get a calendar in the same session.
Which qualification questions matter most?
The three to five answers sales cannot work without: problem, size or volume, timeline, and current tooling. Ask one per turn, give value before each, and skip anything the email domain already tells you.
What metrics should you track for a sales and marketing chatbot?
Capture rate, qualification rate, and meetings booked, in that order. The first ratio that sags identifies the stage to fix. Baseline everything against the form funnel the bot replaced.
Can a chatbot replace sales reps?
No. The bot owns engagement, qualification, and booking around the clock. Sellers own discovery, negotiation, and the relationship. The handoff should be on demand, immediate, and carry the conversation context.
Do chatbots work for lead generation outside business hours?
Yes, and this is their clearest advantage. Buyers research at night and across time zones, and a bot answers now and books the meeting before interest cools, in 95+ languages on SiteGPT. Compare after-hours lead volume before and after deployment to measure the effect.
Sources
- Harvard Business Review, The Short Life of Online Sales Leads, Oldroyd, McElheran and Elkington, March 2011, for the one-hour qualification effect and the audit of 2,241 US companies' lead response times
- diginomica, Salesloft acquires Drift, February 2024, for the Drift acquisition