The chatbot graveyard is full of well-intentioned deployments that lasted three months before being switched off. The pattern is almost always the same: a generic flow that triggers on page load, asks "How can I help?" with four irrelevant quick-reply buttons, and fails to understand anything the visitor actually types.
Open-ended openers with no context perform worse than no chatbot at all. Visitors who get a generic bot feel the company doesn't know who they are or why they're there — and that feeling transfers to the brand.
After analysing 300+ chatbot deployments across UK and GCC businesses, five design traits consistently separate the top performers from the noise:
The best chatbot we ever built opened with: "Are you looking at the ERP for manufacturing or distribution?" It converted at 44% because it was specific. — Lena Kovac, Digital Marketing Lead
The "do I need AI?" question comes up in every chatbot brief. The honest answer: for most SME use cases, a well-designed rule-based bot with smart branching outperforms a naive AI bot because it's predictable, fast, and never hallucinates a product feature.
The only metric that matters is "qualified leads generated per 1,000 visitors". Not session count, not bot opens, not completion rate. Track leads, set a baseline, and optimise everything against it.
A chatbot that captures a lead and emails it to someone is roughly equivalent to a contact form. The delta comes from integration: when a chatbot lead lands directly in your CRM with source, intent, and product interest tagged, your sales team's response time and quality both improve immediately.