chatbots
By Saurav | Founder of saavos | Building in public toward $10k MRR
[!TLDR] Course creators answer the same 10 questions on repeat: "Does this include lifetime access?" "What platform do you use?" "Will this work for beginners?" An AI chatbot trained on your sales page, FAQ, and course outline handles all of them 24/7. Setup is under 30 minutes. The payoff: less support overhead, faster conversion on your sales page, and a better student experience — without a VA or a help desk subscription.
The economics of info products look clean on paper. You build the course once, sell it indefinitely, keep 70–90% margins. The problem nobody talks about: the support overhead doesn't scale down the same way.
Every new cohort brings the same batch of pre-purchase questions from prospective buyers. Every existing student encounters edge cases your course videos didn't fully cover. If you're doing $5K–$50K/month, you're probably spending 3–6 hours per week in email and DM handling questions that are, fundamentally, variations on the same 10–15 FAQs.
That's the problem a course chatbot solves. Not all support — the real coaching, the personalised feedback, the "I'm stuck on this specific concept" moments still need you. But the FAQ overhead can be automated with a $19/month tool and a good afternoon of setup.
To set up a chatbot that actually helps, it's worth separating the question types first.
Pre-purchase questions (from prospective buyers):
These questions live on your sales page or should. If your sales page answered them clearly enough, fewer people would ask. A chatbot here doesn't just reduce support — it converts hesitant buyers who needed one more answer.
Post-enrollment questions (from paying students):
These live in your FAQ, onboarding email sequence, or course platform documentation. The volume is lower per student but accumulates fast with any meaningful cohort size.
A good chatbot handles both categories. A great chatbot separates them — one trained on your sales content for the pre-purchase context, one trained on your student resources for post-enrollment. For most course creators doing under $100K/month, a single well-scoped bot handles both adequately.
The difference between a course chatbot that works and one that frustrates people is almost entirely in source selection.
Train on:
Do not train on:
The clearest mistake I see course creators make: trying to train the bot to teach. A bot that tries to answer "explain constraint theory to me" based on your course transcript will do it badly and undermine your content. Keep the bot scoped to logistics and operations.
This is the setup sequence I'd use for a course business doing $10K–$50K/month.
Step 1 (10 min): Write your master FAQ document. Gather answers to every question you've personally answered more than twice. Format it simply: question on one line, answer in 2–5 sentences on the next. Aim for 20–30 Q&A pairs. This document is your primary training source and it doubles as an FAQ page on your site.
Step 2 (5 min): Add your policies and logistics. Copy in your refund policy, access terms, and community/support information. Add a short platform guide ("to find the bonuses: log in, go to [name], click [thing]"). Keep each section under 300 words.
Step 3 (5 min): Create the bot and write the greeting. Set expectations clearly. Something like: "Hi — I can answer questions about course access, refunds, bonus materials, and what's included. For coaching questions, [reach Saurav at / join the community at]." Don't pretend the bot can help with everything.
Step 4 (5 min): Write the fallback message. When the bot can't answer, it should name the right next step — not just say "I don't know." Example: "I don't have that in our course documentation. For coaching or content questions, post in the community: [link]. For billing and account issues, email support@yourcourse.com."
Step 5 (5 min): Test with 10 real questions. Use questions directly from your inbox — the ones you've answered five or more times. If any come back wrong, find the gap in your training document, add the answer, re-train.
Total: 30 minutes, and you have something real.
Sales page. Add the widget to your course sales page and configure it to appear on scroll depth (50%) or time-on-page (60 seconds). This is where pre-purchase questions happen. A buyer who can't find the refund policy might leave; a chatbot that recites it instantly removes the last friction.
Thank-you / confirmation page. Right after purchase, students have immediate logistical questions: "What happens next?" "Where do I log in?" "When does the live call start?" A chatbot on the confirmation page deflects the bulk of "I just bought — what do I do?" emails.
Course platform FAQ page. If your course platform lets you embed a widget (most do — Teachable, Kajabi, and Podia all support custom HTML), put the bot on your FAQ or help page. Students who are stuck and searching for answers will find the bot before emailing you.
Say you spend 5 hours per week on course support email and DM. At your hourly rate — call it $100 if you're pricing it at founder time — that's $2,000/month in support overhead.
A chatbot that deflects 40% of those questions saves you 2 hours per week, $800/month. Against a $19/month subscription, you're at 40:1 ROI.
The harder-to-quantify piece: conversion. Pre-purchase chatbots on sales pages routinely lift conversion rates 1–3% by answering the questions that were holding buyers back. On a $1,000 course with 100 monthly page visits, a 2% lift is two additional purchases — $2,000. That's not recoverable through email — those buyers bounced before they emailed you.
I want to be direct about this: a course chatbot is not a substitute for real support or real coaching.
If your students are asking for concept explanations, personalized feedback on their work, or accountability coaching — that's the valuable part of what you sell. Do not automate it. Students who pay $500–$2,000 for a course notice when the "coaching" is a chatbot, and it's corrosive to retention and referrals.
Keep the bot scoped to operations: access, logistics, policies, platform navigation. Route everything else to the community, to office hours, to your email. The bot buys back your time on the operational layer so you can give more quality time to the coaching layer.
After launch, set a monthly reminder to review the conversation logs. Look for:
A course chatbot trained once and never touched will drift out of accuracy within 90 days as your policies, platform, or course content evolves. 30 minutes of quarterly maintenance keeps it accurate.
A course sales page has a lot in common with an ecommerce product page — the pre-purchase question pattern is almost identical. AI Chatbot for Ecommerce: Pre-Purchase Q&A That Converts lays out the training hierarchy and conversion math in ecommerce terms, most of which maps directly to info products. If your goal is also capturing leads from prospective students who visit but don't enroll, Lead Capture Chatbot: Turn Site Visitors Into Qualified Emails covers the email capture sequence that works without feeling pushy.
Start free on saavos — paste your FAQ document, write a fallback message, and have your first course chatbot live in under 5 minutes.
Get the next post in your inbox
Honest writing on building, embedding, and shipping AI chatbots. No spam. Unsubscribe anytime.
Two categories: pre-purchase questions from prospective buyers (refund policy, access duration, course format, prerequisites, payment plans, tech requirements) and post-enrollment logistics questions from students (how to access bonus materials, how to join the community, where recordings are posted, platform navigation). The bot is an operational assistant, not a tutor — it handles logistics, not concept explanations. Training it on lesson transcripts to answer content questions produces poor results and risks undermining what you charge for.
Course creators doing $10K–$50K/month commonly spend 3–6 hours per week in email and DM answering repeat questions. The chatbot industry has converged on roughly 40–60% deflection for FAQ-shaped repeat-question volume — we have not validated that on our own customers yet because saavos is pre-revenue. Even at the conservative end (40% deflection, 3 hours/week baseline), that works out to about 1.2 hours/week returned to coaching and content work. The harder number to quantify: pre-purchase conversion lift from answered questions that would otherwise have bounced. A 2% conversion lift on a $1,000 course at 100 monthly visitors is two additional purchases per month.
Train on: your FAQ document (20–30 real questions from your inbox), verbatim refund and access policy, course outline at module-summary level, tech requirements, community and support channel information, and a platform navigation guide. Exclude: lesson transcripts (confuses retrieval), marketing copy and email sequences (too vague), testimonials, and anything about the subject matter itself. A 4,000-word factual FAQ training set is the right target. Lesson transcripts and blog posts are the most common sources of chatbot failure in this category.
Three surfaces in priority order: (1) sales page, triggered at 50% scroll depth or 60 seconds on-page — this is where pre-purchase questions happen and where the conversion lift is highest; (2) thank-you and confirmation page — where new students ask "What happens next?" and "Where do I log in?"; (3) course platform FAQ page — where stuck students are already looking for answers. Most course platforms (Teachable, Kajabi, Podia) support custom HTML embeds. The dashboard or lesson player is lower priority — students there are already engaged.
Four categories need human routing: (1) coaching and concept questions — "I am stuck on this module" needs you, not the bot; (2) billing disputes and refund requests — these have legal and relationship implications that require human judgment; (3) account access failures (the bot cannot fix a broken login); (4) anything emotionally loaded — a student who is frustrated or considering quitting should not be managed by automation. Configure the fallback to route these explicitly: "For coaching questions, post in our community at [link]. For account issues, email [address]."
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.