Most articles about chatbot marketing examples show you screenshots of big brands and leave you wondering what to actually build. This guide does the opposite. First we explain, in plain English, how chatbots capture, qualify and route leads. Then we hand you seven complete chatbot flows with the trigger that fires them, the structure of the conversation script, and the single metric you should watch to know if the flow is working.
Everything below can be rebuilt inside a standard website chat widget, a Messenger or WhatsApp automation, or an AI agent connected to your CRM. No coding knowledge required.
How Do Chatbots Actually Work in Marketing?
A marketing chatbot is not a robot that “talks”. It is a decision tree with a text interface. Every useful marketing bot does the same three jobs, in the same order:
- Capture: it starts a conversation before the visitor leaves, and collects at least one contact point (email, phone, social handle, or a booked slot).
- Qualify: it asks two to four questions that separate a curious browser from a buyer (budget, timeline, company size, use case, location).
- Route: it sends the qualified lead somewhere useful (calendar, sales rep, CRM pipeline stage, email sequence) and sends the unqualified one to self-serve content instead of a human.
That is the whole model. The difference between a bot that generates pipeline and a bot that annoys people is almost never the technology. It is whether these three jobs are clearly separated in the script.
Rule-Based vs AI Chatbots: What Changes for a Marketer
There are two engines behind nearly every example in this article. Many teams now run a hybrid: rules for the qualification and booking part, AI for the open questions.
| Criteria | Rule-based chatbot | AI chatbot (LLM-powered) |
| How it decides | Buttons and if/then branches you write yourself | Understands free text, answers from your content and documents |
| Best for | Lead qualification, demo booking, quote forms, promotions | Pricing questions, product comparison, support deflection, long catalogs |
| Setup time | Hours | Days, mostly spent cleaning the knowledge base |
| Main risk | Dead ends when the visitor has an unexpected question | Confident wrong answers if the source content is thin or outdated |
| Data quality for sales | Very clean (fixed fields) | Messier, needs a structured hand-off step |
Practical rule: use buttons whenever the answer must land in a CRM field, and use AI whenever the visitor is likely to type a question you cannot predict.
The Anatomy of Every Lead Generation Chatbot Flow
Before the examples, memorise this five-block skeleton. All seven flows below are variations of it.
- Block 1: Trigger. Page URL, time on page, scroll depth, exit intent, ad click, or a repeat visit.
- Block 2: Hook. One sentence that names the visitor’s likely intent, not your company. “Comparing pricing plans?” beats “Hi, how can I help you today?”.
- Block 3: Qualifier. Two to four button questions maximum. Every extra question costs completions.
- Block 4: Value exchange. The visitor gets something concrete (a quote range, a calendar, a document, a discount) in return for the contact detail.
- Block 5: Route and confirm. CRM push, email or SMS confirmation, and an owner assigned within minutes, not days.
7 Chatbot Marketing Examples for Lead Generation (With Scripts and Metrics)
1. The Pricing Page Demo Booker
The highest-intent page on most B2B websites is the pricing page. A visitor sitting there for 30 seconds is calculating something. That is the moment to offer a human.
Trigger
Visitor is on /pricing for more than 25 seconds, or scrolls past 60% of the page. Fires once per session, desktop and mobile.
Script structure
- Hook: “Trying to work out which plan fits? I can tell you in two questions.”
- Qualifier 1: “How many people would use it?” Buttons: 1 to 5 / 6 to 25 / 26 to 100 / 100+
- Qualifier 2: “When would you want to be live?” Buttons: This month / This quarter / Just researching
- Branch A (buying window): show a live calendar inside the chat, pre-filtered by the right sales rep. “Here are three slots with a specialist for teams of your size.”
- Branch B (researching): “No problem. Want the plan comparison sheet by email instead?” Capture email, tag as nurture.
- Confirm: summary message plus calendar invite, and a note pushed to the CRM with both answers.
Metric to watch
Meetings booked per 100 pricing page sessions. Ignore chat volume. If this ratio does not move within two weeks, your hook is the problem, not the offer.
2. The Abandoned Quote Request Recovery Bot
Long quote forms leak leads at every field. Instead of shortening the form, put a bot on top of it that catches people as they give up. It is done convincingly by a team that handles this well.
Trigger
Visitor starts the quote form (first field focused) then shows exit intent or stays inactive for 40 seconds without submitting.
Script structure
- Hook: “That form is long, I know. Want me to do it for you in four quick taps?”
- Qualifier 1: service or product category (buttons).
- Qualifier 2: volume, surface, quantity or scope (buttons with ranges, never open text).
- Qualifier 3: deadline (buttons).
- Value exchange: “Based on that, projects like yours usually land between X and Y. Where should I send the detailed quote?” Then ask for one field only: email or phone.
- Route: push to the same pipeline as the form, but tag the source as chat recovery so you can measure incrementality.
Metric to watch
Recovered quote rate: recovered requests divided by abandoned form starts. Anything above 8% is already paying for the widget.
3. The “How Much Does It Cost?” AI Answer Bot
Pricing is the most typed question in almost every chat widget. Refusing to answer it destroys trust. Answering it badly destroys margin. The fix is a bounded AI answer plus an immediate qualifier.
Trigger
Visitor types any message containing price, cost, rate, budget, fees, or “how much”, anywhere on the site.
Script structure
- AI answer, bounded: the bot answers only from an approved pricing document with ranges and the drivers of cost. Example: “Projects start at X. The three things that move the price are scope, integrations and timeline.”
- Immediate qualifier: “Want me to narrow that to your case? Which of these looks closest to your project?” Three or four buttons describing typical packages.
- Value exchange: deliver a tailored range plus an example of a similar project.
- Escalation: “A precise figure needs 10 minutes with a human. Book it here, or leave your email and we send a written estimate today.”
- Guardrail: if the AI has no approved source for the answer, it says so and hands off instead of improvising.
Metric to watch
Containment plus conversion: the percentage of pricing conversations that end either with a satisfied answer or a captured lead. Track the “no answer found” rate separately, it is your content to-do list.
4. The Exit Intent Lead Magnet Bot
A bot beats a pop-up because it asks a question instead of shouting an offer, and because it can adapt the asset to what the person was reading.
Trigger
Mouse leaves the viewport on desktop, or fast upward scroll on mobile, on any blog or resource page. Suppressed for known contacts and for visitors who already engaged in the last 30 days.
Script structure
- Hook tied to the page topic: “Before you go: want the checklist version of this article?”
- Micro-qualifier: “Are you doing this yourself or with a team?” This one answer decides which nurture track they enter.
- Value exchange: deliver the asset in the chat window immediately, then ask for the email to “send a copy you can keep”.
- Soft next step: one button only, “Show me how you would do it for us”, pointing to the booking flow from example 1.
Metric to watch
Email capture rate on exit intent sessions, and then the 30-day assisted conversion rate of those contacts. A lead magnet bot that captures a lot but never assists a sale is collecting collectors.
5. The Paid Traffic Speed-to-Lead Qualifier
If you pay for clicks, the chatbot’s job is to compress the gap between the click and the first human reply. Response speed is one of the strongest predictors of conversion in lead generation, so this flow often returns more than any ad creative change.
Trigger
Landing page visit with a paid UTM parameter. Bot opens automatically after 5 seconds with the offer from the ad repeated word for word.
Script structure
- Hook matching the ad: “You clicked on the free site audit. Want it done this week?”
- Qualifier 1: disqualifier first. “Which best describes you?” with one option that politely exits non-buyers (student, job seeker, agency reselling).
- Qualifier 2: budget or size band, in ranges.
- Live routing: during business hours, ping the rep on duty and show “Someone is joining in under a minute”. Outside hours, offer the calendar.
- Backup capture: always take the phone or email before attempting the live hand-off, so a failed transfer is still a lead.
Metric to watch
Median time to first human reply, plus cost per qualified conversation. Compare it against cost per form fill from the same campaign, that is the only fair benchmark.
6. The After-Hours and Weekend Booking Bot
Depending on the industry, a large share of website traffic arrives when nobody is at the desk. Those visitors do not wait, they open the next tab and contact a competitor. There is more on it in Best Chatbot Examples for Businesses from Leading Brands.
Trigger
Any session outside your opening hours, on contact, service or product pages.
Script structure
- Honest hook: “The team is offline until 9am. I can still get you on the list, first slots are tomorrow morning.”
- Qualifier: reason for contact (buttons: new project / existing order / support / partnership). Support and partnership are routed away from sales instantly.
- Value exchange: show real available slots for the next working day, or take a callback request with a time window the visitor chooses.
- Expectation setting: confirm by email or SMS with the exact promised time. Never write “as soon as possible”.
Metric to watch
Share of total leads captured outside business hours, and the show-up rate on those bookings. A low show-up rate usually means the confirmation message is missing or too vague.
7. The Returning Visitor Re-Engagement Bot
Someone on their third visit who has not converted needs a different conversation than a first-time reader. This is the flow most companies never build, and it is often the cheapest win. Related reading: The Complete Guide to Chatbot Marketing.
Trigger
Third or later session within 30 days, no form submitted, no meeting booked.
Script structure
- Hook that acknowledges the pattern: “You have been comparing options for a while. What is still unclear?”
- Objection menu: buttons naming the real blockers. Price / timing / integration / need internal approval / not sure you are the right fit.
- Tailored proof per objection: a case study for fit, a payment option for price, a one-page internal business case for approval, a technical sheet for integration.
- Low-commitment ask: “Want a 15-minute call with no deck, just your questions?”
- Data value: log the chosen objection in the CRM. After 50 conversations you have a ranked list of what blocks your sales, which is worth more than the leads themselves.
Metric to watch
Objection distribution first, then conversion rate of returning-visitor conversations. Use the distribution to rewrite your landing pages.
Which Metric Belongs to Which Flow
Chat volume is a vanity metric. Match each flow to one number and review it weekly.
| Flow | Primary metric | Warning sign |
| Pricing page demo booker | Meetings per 100 pricing sessions | High opens, no bookings |
| Quote recovery | Recovered quotes / abandoned starts | Drop-off on question 3 |
| Pricing answer bot | Containment + lead capture rate | Rising “no answer found” rate |
| Exit intent magnet | Email capture rate, 30-day assist | Emails that never open anything |
| Paid traffic qualifier | Time to first human reply | Transfers that time out |
| After-hours booking | Off-hours lead share, show-up rate | No-shows above 40% |
| Returning visitor bot | Objection distribution | Everyone picks “other” |
Five Mistakes That Kill Chatbot Conversion Rates
- Opening with “How can I help you?” It forces the visitor to do the thinking. Name their intent based on the page instead.
- Asking for the email before giving anything. The value exchange has to come first, even if the value is just a price range.
- Too many questions. Every question past the fourth cuts completion. Let the sales call collect the rest.
- No visible exit. Always offer “I would rather browse” and remember the choice for the session. Interruption without an exit is a bounce machine.
- No owner after capture. A bot that fills a database nobody checks is worse than no bot, because the visitor believed they were being taken care of.
How to Launch Your First Flow This Week
- Pick the single page with the highest commercial intent and decent traffic (usually pricing, a service page, or a paid landing page).
- Write the three qualification questions your sales team asks on every first call. Turn them into buttons with ranges.
- Decide the value exchange: calendar, price range, document, or callback window.
- Connect the output to your CRM with a dedicated source tag, and assign an owner with a response deadline.
- Run it for two weeks untouched, then change one thing at a time: first the hook, then the number of questions, then the offer.
- Only once the rule-based version converts, layer AI on top for open questions. AI amplifies a good flow, it does not fix a bad one.
Need help mapping these flows to your funnel and wiring them into your CRM? That is exactly the kind of work we do at adhurl.com, and it usually starts with one page, one flow and one metric.
Frequently Asked Questions
What are some good examples of chatbots in marketing?
The most reliable examples are not novelty campaigns, they are repeatable lead flows: a pricing page demo booker, a quote form recovery bot, a bounded pricing answer bot, an exit intent lead magnet, a paid traffic qualifier, an after-hours booking assistant, and a returning visitor objection handler. All seven are detailed above with their triggers and scripts.
What are the four types of chatbots?
Most classifications land on: menu or button-based bots, keyword recognition bots, rule-based (decision tree) bots, and AI or generative bots that understand free text. Hybrid bots that mix a rule-based qualification path with an AI answer layer are now the standard setup for marketing teams.
Do chatbots still work now that people are used to AI assistants?
Yes, but expectations are higher. Visitors accustomed to AI assistants will type full questions instead of clicking buttons, and they abandon instantly when the bot loops. The winning pattern is fewer questions, honest answers including price ranges, and a fast route to a human.
How many questions should a lead qualification chatbot ask?
Two to four. Ask only what changes the routing decision or the price. Everything else belongs on the discovery call or in a post-booking email.
Should the chatbot answer pricing questions?
Give a range and explain the three drivers behind it. Silence on price is read as “expensive and evasive” and pushes visitors to competitors. A range plus a qualifier keeps the conversation alive without committing you to a figure.
What is a good conversion rate for a marketing chatbot?
It depends on placement, so benchmark against yourself. As a starting point: on a high-intent page, expect a meaningful share of engaged conversations to end in a captured contact, and on abandoned form recovery, anything above 8% recovered is worth keeping. Always compare the chatbot against the form on the same page rather than against industry averages.
Do chatbots hurt SEO?
Not the conversation itself, but a heavy widget that blocks rendering or a popup that covers content on mobile can hurt both page experience and conversions. Load the widget after the main content, keep it small on mobile, and never cover the primary call to action.