A Support Chatbot for a Solo Business — Deflect the Repeat 60% Without Sounding Robotic
A knowledge-base chatbot is the closest thing a solo business has to a support hire that works nights and weekends: it answers the repeat 60% instantly from your own documentation, collects the rest as structured tickets, and never invents a refund policy at 3am — if you scope it correctly.
This guide covers the weekend deployment: KB preparation, bot scoping, the never-automate list, and the transcript-review habit that keeps answers trustworthy.
The short answer
- KB-trained chatbots resolve 40-70% of common questions on well-documented products and services.
- Instant answers at any hour measurably lift customer satisfaction even when the human alternative was only hours away.
- Bot failures cluster in edge cases — the scoping decision (what it answers vs routes) determines satisfaction more than model quality.
Who this playbook is for
Built for solo founders drowning in repeat customer questions who cannot hire support staff.
Step 1: Prepare the knowledge base first — the bot is only as good as it
Fifteen to twenty articles covering your top questions, written plainly with steps and screenshots. This is the actual work of the project; the bot is plumbing on top. If you already run the AI support triage workflow, the KB exists — connect it and skip ahead.
Step 2: Pick a KB-grounded tool, not a free-form one
Tools: Intercom Fin, Chatbase, Crisp AI — products that answer only from your documents and cite them. Avoid free-form chatbots that answer "creatively": a support bot’s job is accuracy from your source of truth, and grounding is the feature that provides it.
Step 3: Scope what it answers — and what it must never touch
Bot answers: how-to, access, pricing explanations, status, policies as documented. Bot never answers: refunds, custom quotes, legal or data questions, angry escalations — those route to you as tickets with context attached. Write the never-list publicly in the bot’s personality so handoffs feel designed, not evasive.
Step 4: Design the graceful handoff
Every bot reply ends with an escape hatch: "Was that helpful? Talk to a human →". Low-confidence answers route automatically with the conversation transcript attached. A handoff that preserves context is the difference between customers feeling helped and feeling bounced — transcript transfer is non-negotiable.
Step 5: Review transcripts weekly and feed misses back
Fifteen minutes weekly: scan conversations for wrong answers, confusions, and questions the bot dodged. Every miss becomes either a KB update or a scoping fix. This review loop is the entire maintenance cost of the system — skip it and accuracy decays silently until a public complaint tells you.
Your weekly operating rhythm
| Day | Action | Time |
|---|---|---|
| Weekend | KB check; bot configured and scoped; handoff tested | 1 day |
| Daily | Ticket queue for routed conversations | 20 min |
| Weekly | Transcript review; KB and scope updates | 15 min |
| Monthly | Deflection and satisfaction review | 30 min |
KPIs that tell you it is working
| Metric | Healthy target | Why it matters |
|---|---|---|
| Deflection rate | 40-70% of chats | The share your KB absorbed without you |
| Handoff rate with context | 100% | Every human-bound chat arrives with its transcript |
| Wrong-answer incidents | Trending to zero | The weekly review’s output metric |
| After-hours resolution share | Tracked and celebrated | The quiet superpower of the deployment |
Common mistakes to avoid
- Letting the bot freestyle on policies. "The bot said I could refund" is a self-inflicted wound; grounding plus the never-list prevents it.
- Shipping with a thin KB. Ten articles deflect ten questions; the ROI of writing the other ten first is the whole business case for the bot.
- Hiding the human option. Bots that trap customers generate rage-reads on review sites; a visible escape hatch generates patience.
A tool stack that fits a one-person budget
| Tool | Where it fits |
|---|---|
| Chatbase / Crisp | Affordable KB-grounded bots |
| Intercom Fin | The premium option with strong resolution analytics |
| Notion / Helpjar | The KB source the bot cites |
| Your help desk | Where routed conversations land with context |
Keep going
Use these internal references while implementing this guide:
- One Person Company Hub
- How to Start a One Person Company
- Solopreneur Operating System
- n8n for Solopreneurs
- Your AI Chief of Staff
- A Prompt Library That Runs Your Business
FAQ
Q: Will a chatbot make my business feel small or cheap?
Executed well, the opposite: instant accurate answers at midnight read as competence, not cheapness. What reads as cheap is a bot that fails and hides the human option. Grounding, scoping and handoff design are what keep the perception premium.
Q: How much does this cost?
From free tiers (Crisp, Chatbase starters) to $99+/month for Fin at volume. Most solo businesses run a capable KB bot for under $50/month — compare against even one hour of your evening per day and the math is rarely close.
Q: Can the bot also do sales questions?
Pricing explanations and fit questions, yes — from your pages. Custom quotes, negotiations and "will it work for my case?" stay human: sales answers carry commitment, and commitment stays with the founder. Some bots book calls, which is a perfect handoff.
Q: How do I measure if it’s actually working?
Three numbers: deflection rate (share answered without you), wrong-answer incidents (trending down), and after-hours resolutions (pure bonus). If deflection is high but complaints appear, weight accuracy over volume — one wrong refund answer costs more than fifty right ones.
Get the weekly operating brief
Every Monday: 3 moves, 5 minutes. Actionable strategy for your one-person company — no fluff, no filler.