Test Your Chatbot

Send test messages, review answers, and iterate before going live.

Before you share your WhatsApp number or embed the widget on your live site, run through a real conversation. This catches gaps in your knowledge base and tunes the AI's tone for your business.

1. Test in the web widget

Open the Settings page in the dashboard. You'll find an Embed Code section with two snippets you can copy:

  • Floating bubble script — a self-contained <script> tag that adds a chat button to the bottom-right of every page. Paste it once before the closing </body> of your site.
  • Iframe — a fixed-position <iframe> if you prefer to control placement yourself.

For testing without embedding, open a new browser tab and visit www.ozcenklabs.com/chat/<your-klinik-id> — the same chat interface appears as a full page. Your klinik id is shown in the Settings page.

Try these conversations:

  • Information request — "Do you offer X service? How much does it cost?"
  • Appointment booking — "Can I book for tomorrow at 3pm?"
  • Cancellation — "I need to cancel my Friday appointment."
  • Off-topic — "What's the weather like?" (the AI should politely redirect to your business.)
  • Human handoff — "Can I speak to someone?" (the AI should offer to flag the conversation.)

2. Test on WhatsApp

After connecting your WhatsApp number (see Connect WhatsApp Business), send a message to your own number from a different phone. The AI replies within seconds.

Cover the same scenarios as the web widget. WhatsApp behaves identically — same AI, same knowledge base — but it's worth confirming end-to-end delivery.

3. Review every conversation

Go to Live Chats in the dashboard. Each test thread appears with:

  • Full message history (incoming + AI replies).
  • Captured contact data (name, phone, request).
  • Linked lead and appointment, if either was created.
  • A handoff toggle to take over the chat manually.

Read every reply and ask yourself:

  • Is the information correct? If a price is wrong, your knowledge base needs an update.
  • Does the tone match your business? Sector tone is configured globally; if it feels off, email hello@ozcenklabs.com.
  • Did the AI capture leads correctly? Names, phone numbers, and the requested service should appear in the Leads module.
  • Did appointment booking work? Check Appointments for new entries.

4. Try edge cases

Stress-test before launch:

  • Outside working hours. The AI mentions your hours and offers to take a message. Configure hours in Settings → Working Hours.
  • Different languages. The AI replies in whatever language the customer writes in (English, Turkish, Russian, German, Arabic, more).
  • Very specific questions. If the answer isn't in your knowledge base, the AI should admit it and offer to take a message or hand off.
  • Long conversations. The AI keeps the last 10 messages in context. Verify it remembers what was said earlier.

5. Hand off to a human

Some conversations need a person. Customer asks for a manager, complains, or has a sensitive issue — the AI flags the contact for handoff and stops auto-replying. From the Live Chats view you can:

  • Type a manual reply that goes out as your number.
  • Toggle handoff back off when the issue is resolved — AI auto-replies resume.

6. Tune and iterate

If something is off, fix the source:

SymptomFix
Wrong price quotedUpdate the price in your catalog (Services / Products / Portfolio / Vehicles)
Service not recognisedAdd it to your catalog or AI Knowledge Base
Wrong working hoursUpdate Working Hours in Settings
AI invented a factAdd the correct fact to the knowledge base; AI prefers your data
AI too formal/casualEmail us — sector tone is configured at the platform level

7. Go live

When you're happy with the answers:

  1. Paste one of the embed snippets from Settings into your website.
  2. Share your WhatsApp number on your social profiles, business cards, and signage.
  3. Watch the Leads module — every captured customer appears with full context.

Need help?

Email hello@ozcenklabs.com or call +44 7719 750785. We answer within one business day and we're happy to look at specific conversations with you.

Test Your Chatbot — OzcenkLabs Docs | OzcenkLabs