The Honest Range
An AI chatbot for an SMB can cost anywhere from $99/month to $250,000 one-time. The price reflects what's actually being built. Most SMBs buy the wrong thing for their actual need.
Here is the breakdown of what each price tier delivers.
Tier 1: Off-the-Shelf SaaS Chatbot ($40-$200/month)
Examples: Intercom Fin, Drift AI, Tidio AI, HubSpot AI Chatbot, basic Chatbase setups.
What you get:
- A widget you embed on your website
- An LLM (usually GPT-4 or Claude) trained on your help docs
- Basic conversation flow
- Some integration with your CRM
What it handles well:
- Tier-1 FAQ deflection
- Hours, location, basic pricing questions
- Routing serious questions to humans
What it does poorly:
- Anything that requires writing to your systems
- Anything that depends on real-time data
- Anything specific to your industry workflow
- Multi-turn conversations with state
When this is the right buy: You have <50 employees, your support volume is <100 tickets/week, and 80%+ of incoming questions are FAQ-style.
Tier 2: Custom Build, Simple Scope ($5K-$15K one-time, ~$500/month operations)
What this typically includes:
- Custom chatbot trained on your specific docs, products, or knowledge base
- One or two key integrations (e.g., your CRM + your scheduling tool)
- Specific personality matched to your brand voice
- Basic analytics
When this is the right buy: Your business has unusual workflows or jargon, off-the-shelf bots gave wrong answers in your test, and you have one specific use case you want automated.
Tier 3: Custom Build, Production Scope ($25K-$85K one-time, ~$2K-$5K/month operations)
This is the modal production-scope chatbot engagement for SMBs. It typically includes:
- RAG over multiple data sources (docs, CRM data, product catalog, historical tickets)
- Multi-channel deployment (web, WhatsApp, SMS, Slack)
- Integration with 3-5 systems (CRM, calendar, billing, ticketing, knowledge base)
- Eval suite that runs continuously
- Cost engineering (per-conversation dollar tracking)
- Custom escalation logic (when to hand off to humans, with what context)
- Production monitoring + ongoing operations
When this is the right buy: The bot is genuinely doing work, not just answering FAQs. It needs to write to systems. The volume is meaningful (~500+ conversations/month). You have a target ROI (X% deflection, Y hours saved).
Tier 4: Enterprise / Multi-Region ($100K-$250K+ one-time)
Multi-language, multi-region, multi-brand, regulated industry (HIPAA, financial services), or integrated into 10+ systems.
If you're not sure whether you're at this tier, you're probably not.
What ROI Actually Looks Like
For a Tier 3 build at $25K-$85K, the ROI math we see most often:
Customer support deflection scenario:
- Pre-build: 500 tickets/month at avg 12 min handling time = 100 hours/month
- Post-build: 35% deflection rate (industry realistic for well-built bots)
- Savings: 35 hours/month at $25/hr loaded cost = $875/month direct savings
- Plus faster response on the 65% that escalate (customer satisfaction lift)
- Payback: 8-12 months on a $25K build
Lead qualification scenario:
- Pre-build: 200 inbound leads/month, response time 4-12 hours, conversion 12%
- Post-build: Response time <1 min, conversion 18%
- Revenue lift on a $500 avg deal: $6,000/month additional revenue
- Payback: 4-6 months on a $25K build
The lead qualification case is usually the highest ROI. SMBs underestimate how much they lose by responding slowly.
What Most SMBs Get Wrong
- They buy Tier 1 when they need Tier 2 or 3. The cheap chatbot frustrates customers because it can't actually do anything. They cancel after 60 days.
- They buy Tier 3 when they need Tier 1. They overspend on capabilities they won't use, then can't show ROI.
- They skip the eval suite. The bot ships, looks good, then silently degrades over 60 days as customers ask edge cases that weren't tested.
- They forget the operations cost. A production chatbot needs monthly attention. Model updates, prompt tuning, new product changes that need to be added to the RAG. Budget ~$2-5K/month for any meaningful production bot.
- They build without integration. A chatbot that can't write to your CRM, schedule on your calendar, or check inventory in your stock system is a FAQ tool, not an AI agent.
How to Scope Yours
Start with an AI Readiness Audit. One week. You'll come out with:
- Workflow audit showing where AI fits in your business
- Top 3 use cases ranked by ROI
- A working demo of one of them on your actual data
- Honest recommendation on which tier you need
If the audit recommends Tier 1, that's not bad news. It saves you $20-80K and a year of regret.
If the audit recommends Tier 3, the audit cost credits to the engagement if you continue.
The wrong buy is the most expensive mistake. The audit fixes that.