Playground Testing Before Go-Live

What this guide covers

This guide explains how to use the chatbot playground to test quality before go-live. You will run prompt scenarios, verify desktop/mobile behaviour, clear and retest conversations, and complete a release-ready sign-off check.

Before you start

• Select the correct tour location in Chatbot.
• Ensure chatbot configuration exists and is Active.
• Complete baseline setup (information, documents, triggers, and customisation) before final testing.

1) Open playground and prepare the test context

1. Go to Chatbot > Playground.
2. Confirm the selected tour location is correct.
3. Verify the playground loads without configuration errors.
4. Keep a simple test log for pass/fail notes.

The playground uses live API behaviour, so test results are operationally meaningful.

2) Run scenario prompts

Use the built-in prompt categories to test common visitor intent:

• Tour overview
• Model navigation
• Bookings and availability
• Opening hours
• Amenities
• Directions and support

Also add your own edge-case prompts for ambiguous or incomplete user questions.

3) Validate desktop and mobile behaviour

1. Test in Desktop mode.
2. Switch to Mobile mode and rerun key prompts.
3. Compare response clarity, readability, and interaction flow in both contexts.
4. Confirm widget open/close behaviour is stable.

Do not assume desktop outcomes will match mobile outcomes.

4) Reset and retest cleanly

1. Use Clear Chat between test sets.
2. Re-run priority prompts after each major config change.
3. Confirm responses are consistent across repeated runs.

This isolates test sessions and avoids false confidence from prior context.

5) Final go-live sign-off

Before publishing, confirm:

• High-priority prompts produce correct responses.
• Triggered behaviours operate as expected.
• Message quality is clear, concise, and on-brand.
• No critical failures remain in desktop or mobile mode.
• Test log is complete and approved for launch.

Common issues

• Playground not available: select a tour and activate chatbot configuration first.
• Responses are generic or weak: revisit instructions, information sections, and documents.
• Different behaviour between desktop and mobile: review customisation and rerun both modes.
• Inconsistent outputs between runs: clear chat and re-test with controlled prompts.
• Configuration error state: resolve setup issues in Settings, then return to playground.

Validation checklist

• Correct tour location selected.
• Chatbot is active.
• Prompt scenarios tested across key user intents.
• Desktop and mobile checks both passed.
• Final release sign-off recorded.

Final note

Use playground testing as a formal pre-release gate, not a quick visual check. Consistent scenario testing and clear sign-off criteria reduce launch risk and improve chatbot quality in production.