Replacing a Failing Rule-Based Chatbot with AI Assistant
TL;DR
An AI assistant replaced a failing rule-based chatbot turning low-quality traffic into qualified revenue conversations.
Impact
Engagement increased from 1.4% → 2.3%
Chat → MQL conversion increased 1.2% → 6.8% (5.6×)
MQL → Opportunity conversion increased 10.4% → 15% (44%)
Why This Mattered
The existing chatbot created friction instead of clarity. It misrouted users, failed to answer real compliance questions, and generated low-intent leads that wasted sales time.
User: Prospects needed fast, credible answers about international hiring — not generic form submissions.
Sales: Low-quality MQLs reduced efficiency and eroded trust in marketing channels.
Business: Compliance is Velocity Global’s differentiator. Introducing AI had to increase conversion without compromising legal accuracy across international employment law.
What Shipped
A legally constrained AI assistant embedded directly into the marketing site — designed to increase qualified engagement without introducing compliance risk.
Shipped Experience
Mobile-first conversational AI embedded in-site
Guided question starters (pricing, compliance, EOR education)
Progressive value reveal before email unlock
AI responses constrained within predefined compliance buckets
Fast Follows:
Full-screen mode
Entry-point A/B testing and optimization
Response rating system for quality feedback
Initial Launch (Executive-Directed)
High-visibility fixed entry
Optimized Variant (Post A/B Test)
Contextual motion-based trigger
Key Decision
Design for credibility over spectacle.
Rather than creating a highly persuasive, attention-grabbing AI layer, I constrained the assistant:
Limited question starters to reduce cognitive load
Structured conversations within approved compliance buckets
Embedded source references (“receipts”) where applicable
Prioritized clarity and trust over novelty
Simultaneously, I accepted the executive-directed high-visibility entry point to protect the timeline but prepared an alternative motion-driven interaction for immediate A/B testing.
Outcome
The AI assistant shifted from a novelty experiment to a measurable revenue driver.
64%
Increase in chatbot engagement
5x
Increase in Chat → MQL conversion
44%
Increase in MQL → opportunity
In Closing
This project shifted AI from a novelty experiment to a revenue-driving system.
It proved that AI in regulated environments can be both growth-oriented and legally constrained when designed intentionally.
Role
Sole Product Designer • Led conversational system design, guardrails, and launch execution
Delivered MVP in < 1 quarter under strict legal constraints
Impact Type
Revenue growth through improved lead generation and conversion
Team
PM, Eng (6), BenOps, Legal
