chatbots
By Saurav | Founder of saavos | Building in public toward $10k MRR
[!TLDR] A lead capture chatbot converts 2–5% of site visitors into email leads — 3–4× the rate of a static contact form — by starting a conversation instead of asking for commitment upfront. The setup takes under an hour: train the bot on your product positioning, configure a qualification question sequence, and connect to your CRM or email list. The visitors who engage are already pre-qualified by intent; the bot's job is to capture the email before they leave and hand off context to sales.
Most B2B sites convert 0.5–1.5% of visitors into form submissions. That's the industry average. And it's mostly a psychological problem, not a design problem.
A contact form is a commitment request. It says: give me your email address, your company name, your job title, describe your use case, and wait for someone to contact you in 24 hours. For a visitor who's still deciding whether your product is relevant, that's too much friction for too little certainty.
A lead capture chatbot flips the sequence. The visitor asks a question. The bot answers it — useful, specific, cited from your actual product content. Somewhere in that conversation, when the visitor is already engaged, the bot asks: "Want us to send you the comparison doc?" or "Can I have your email so we can follow up with the pricing breakdown?"
The ask comes after value, not before. That's why conversion is 3–4× higher.
Not every visitor who engages with your chatbot is the same. Understanding which type you're capturing changes how you configure the qualification logic.
Intent-high, information-lacking. This visitor knows they want a solution like yours. They're asking comparison questions ("How does this compare to Intercom?"), pricing questions ("What's included in the Pro plan?"), or integration questions ("Does this work with Salesforce?"). These are your best leads. The bot should answer the question directly, then offer to send detailed documentation or book a demo. Conversion rate on this segment is often 8–15%.
Intent-medium, exploring. This visitor is researching the problem space. They ask "How do AI chatbots work?" or "What's the ROI of a support bot?" They're not ready to buy today but they're in your funnel. Capture their email with a content offer ("I can send you the ROI calculator we use with customers") rather than a sales ask. Nurture over weeks, not days.
Intent-unknown, browsing. This visitor may have landed from a blog post or social link and isn't clearly in-market. Don't push for an email too early — it reads as aggressive and lowers trust. Let them engage first. If the conversation goes two or more turns, light intent is showing. That's when to make an offer.
Your chatbot should handle all three gracefully. The mistake is configuring a single aggressive capture prompt that fires on every visitor regardless of engagement depth.
Lead capture bots have a different source requirement than support bots.
A support bot is trained on docs and FAQs — operational content for existing customers. A lead capture bot is trained on your best sales content: the pages that communicate value, differentiation, and fit.
Good sources for a lead capture chatbot:
What to exclude:
The visitor asking "How long does setup take?" needs the honest answer: five minutes to train, one script tag to deploy. They don't need a blog post about RAG architecture. Tight sources produce confident, specific answers that move people from curious to captured.
Once you have the bot trained on your sales content, you need a conversation flow that captures the email at the right moment.
Here's the pattern that works:
Answer the question first. Whatever the visitor asks, answer it well. This is the trust-building moment. A vague or evasive answer here ends the conversation.
Follow up with one qualifier. After the answer, ask a single relevant question: "What kind of site are you looking to add this to?" or "Are you evaluating for yourself or a team?" This surfaces context that makes the handoff to sales useful.
Make the offer. "Based on what you've described, I can send you the exact pricing breakdown for your use case — what's the best email to reach you?"
On capture, set an expectation. "Got it. You'll hear from Saurav within a few hours with the breakdown and a few relevant examples."
The sequence works because it creates a transaction: the visitor gives an email in exchange for something specific they already indicated they want. The email isn't a toll; it's a delivery address.
One common mistake: asking for more than an email at capture time. Name, company, phone number — each additional field cuts conversion by 10–20%. Capture the email. Get the rest in the follow-up.
A chatbot that captures emails but has no downstream workflow is a dead end.
The integrations worth setting up before you go live:
Email platform (Mailchimp, ConvertKit, ActiveCampaign). Map the captured email to your lead nurture sequence immediately. The chatbot conversation context — what they asked, what segment they're in — should pass as tags or custom fields so the follow-up email is relevant, not generic.
CRM (HubSpot, Salesforce, Pipedrive). For B2B SaaS with any kind of sales motion, create a contact record with the full conversation log as a note. The sales rep who follows up needs to know the visitor asked about Salesforce integration and $200/month pricing — not just that someone submitted a form.
Slack or email notification. For high-intent signals (comparison questions, pricing questions, "book a demo" intent), fire a Slack message to the sales channel immediately. Real-time notification on hot leads matters more than overnight batch delivery.
Most modern chatbot platforms — including saavos — connect to these systems via native integrations or Zapier. Set up the CRM connection before you launch; going back to enrich records retroactively is painful.
Chat engagement rate. 5–15% of visitors engage with the chatbot. Below 5% suggests the launcher isn't visible or the opening message doesn't invite engagement. Above 15% can indicate the launcher is too intrusive.
Email capture rate (per engaged visitor). 15–35% of visitors who engage with two or more turns should convert to an email capture. Below 15% means the qualification sequence or offer needs work. Above 35% is possible on high-intent pages like pricing or comparison pages.
Lead quality rate. Of the emails captured, how many are relevant to your ICP? Track by measuring open rate on the follow-up, reply rate, and conversion to demo or trial. A lead capture rate of 30% is only good if the leads are real — if 90% bounce the follow-up email, the offer was too broad.
Time-to-follow-up. Leads that get a follow-up within one hour convert at 7× the rate of leads followed up the next day. The chatbot buys you the conversation; your response speed closes it.
Your highest-intent pages are where you want the bot most visible.
Pricing page. Visitors here are evaluating. A chatbot that can answer "What's included in the plan?" or "Can I start on free?" removes the last friction before a sign-up or contact form.
Comparison and alternatives pages. If you have a "vs Competitor" page, the visitor is already in buy mode. The chatbot here should be proactive — triggered on scroll depth or time on page, not just on click.
Demo request page. Replace or supplement the static form. A chatbot that pre-qualifies before the calendar link appears converts better and wastes less demo capacity on poor-fit leads.
Blog posts with commercial intent. "Best AI chatbot for SaaS" posts attract buyers. A chatbot that appears on these posts at high scroll depth catches visitors who made it through your content and still want more.
The homepage is worth covering too, but it's not your best surface for lead capture — the visitor intent is too varied. Focus placement energy on the pages where visitors are already in evaluation mode.
You don't need a sophisticated multi-step flow on day one. Start with a chatbot trained on your three best sales pages, a simple email capture offer, and a Slack notification for high-intent conversations.
Review the first 50 conversations. You'll see quickly which questions signal strong intent and where the conversation falls off. Tune from there.
If your site is on Shopify, the setup path has some platform-specific steps — see How to Add a Chatbot to Your Shopify Store in 2026 for the exact embed sequence and where to place the widget on product vs. cart pages. For course creators and info product sellers, the lead capture pattern maps closely to the pre-enrollment context covered in AI Chatbot for Online Courses: Turn FAQ Overhead Into Auto-Answers.
Start free on saavos — train on your product pages in five minutes, capture your first lead the same day.
Get the next post in your inbox
Honest writing on building, embedding, and shipping AI chatbots. No spam. Unsubscribe anytime.
A lead capture chatbot converts 2–5% of site visitors into email leads, versus 0.5–1.5% for a static contact form — roughly 3–4× higher. The difference is sequencing: a chatbot delivers value first (answers a real question) and asks for the email after the visitor is already engaged. The form asks for commitment upfront with no demonstrated value. Visitors who engage with two or more turns of conversation convert to email capture at 15–35% of that engaged subset.
Train on your best sales content, not your support docs. Include: product overview and feature pages, pricing with plan comparison, comparison pages ("vs Competitor X"), customer use cases with concrete numbers, integration list, and how-it-works / setup time explanation. Exclude blog archives, detailed technical docs, and press content. A visitor asking "How long does setup take?" needs the honest answer, not a RAG hit on a blog post about architecture.
Three-step sequence: answer the question first, ask one qualifying question ("What kind of site are you adding this to?"), then make a specific offer ("I can send you the pricing breakdown for that use case — what email should I use?"). The ask comes after demonstrated value, not before. On capture, set an expectation for when and how they will hear from you. Never ask for more than an email at capture time — each additional field (name, company, phone) cuts conversion 10–20%.
Prioritize high-intent pages: pricing page (visitors are evaluating), comparison and alternatives pages (visitors are in buy mode), demo request page (supplement or replace the static form), and blog posts with commercial intent (high scroll depth signals readiness). The homepage is worth covering but has mixed intent — focus energy on pages where visitors are already in evaluation mode. Trigger the chatbot proactively on high-intent pages at scroll depth or time-on-page rather than waiting for the visitor to click the launcher.
Three integrations matter: (1) email platform (Mailchimp, ConvertKit, ActiveCampaign) — map the captured email to a lead nurture sequence immediately with conversation context as tags; (2) CRM (HubSpot, Salesforce, Pipedrive) — create a contact record with the full conversation log as a note so sales knows what was asked; (3) Slack notification — fire a real-time alert to the sales channel for high-intent signals like comparison questions or pricing asks. Set up integrations before launch; retroactive data enrichment is painful.
Builds tools for solopreneurs and small SaaS teams who don't have an afternoon to spare.
Paste your URL. Train your bot. Drop one script tag. No credit card.