How Do You Stop Losing Conversations in Zendesk Sunshine Conversations?

To stop losing Sunshine Conversations when customers ignore your bot's buttons, turn on AI agent conversations as tickets in Admin Center (Channels > Messaging and social > Messaging > Manage settings), then add the condition Ticket > Support type | Is | AI agent to a view so agents can see them. If you need a real, editable ticket for every abandoned chat, you have to build it yourself with a conversations integration webhook and a small piece of middleware, which needs Suite Professional or above.

The problem is common with custom bots: a customer sends a first message, the bot offers buttons, and the customer closes the chat without clicking anything. That conversation never reaches an agent, and nobody follows up.

Why do Sunshine Conversations disappear when customers skip the buttons?

When a customer sends their first message in Sunshine Conversations, your custom Zendesk bot might reply with options like "Get a Price Quote" or "Technical Support." If the customer doesn't click a button (they got distracted, closed the chat, or didn't want to use the buttons), the conversation stays inside Sunshine Conversations and agents never see it.

As one Zendesk user put it: "These conversations at in Sunshine Conversations only, completely invisible to agents and lost revenue."

The root cause is that a bot conversation doesn't become a normal ticket until something hands it to an agent. The data exists, but reading it means calling the API, which isn't practical for daily work. Every one of those lost chats is a customer who asked for help and didn't get it.

How do AI agent tickets fix it?

The most straightforward fix is Zendesk's AI agent tickets feature. It creates a visible, trackable ticket for conversations handled entirely by an AI agent.

How AI agent tickets work

Zendesk creates an AI agent ticket when a customer begins a conversation with a messaging or voice AI agent, or just types in the Web Widget. The ticket stays read-only while only the AI agent is involved. Once the conversation is escalated to a human, it stops being an AI agent ticket and becomes a regular, editable ticket.

What you get

• Visibility: bot conversations show up as tickets in Agent Workspace once your views include them
• An audit trail: read the full conversation between your AI agent and the customer
• Resolution checks: look through solved tickets to confirm the AI agent resolved them properly
• Customer context: AI agent tickets appear in the customer's interaction history, so agents see them next time

How do you turn on AI agent tickets?

In Admin Center, click Channels in the sidebar, then select Messaging and social > Messaging. Click Manage settings, expand the AI agent conversations as tickets in Agent Workspace section, select Turn on AI agent conversations as tickets, and click Save.

Why can't agents see them after you turn them on?

AI agent tickets don't appear in existing views by default. Zendesk added the condition Ticket > Support type | Is | Agent to views created before the feature reached your account, so those views filter them out. Add Ticket > Support type | Is | AI agent to a view for AI agent tickets only, or Ticket > Support type | Is | All to see both kinds. AI agent tickets also stay out of Agent Home, Explore reports (except the Resolution type field) and the API endpoints that list tickets.

To search for AI agent tickets in Agent Workspace:

1. Perform a search
2. Select the Tickets filter
3. Select Support type
4. Select AI agent

Source: Zendesk Help Center - Understanding and viewing AI agent tickets and Finding AI agent tickets in your account

Can a trigger create a ticket when a customer doesn't click a button?

Not on its own. Zendesk's triggers can act when a conversation reaches an agent, but none of them fires on "the customer went quiet." For that you need a timeout outside Zendesk.

How does the timeout approach work?

The method is a custom integration with a timeout fallback:

1. When a customer starts a conversation, wait 30-60 seconds
2. If the customer doesn't click a button within that time, create the ticket automatically
3. No conversation falls through the cracks

According to community feedback, this needs custom middleware: a webhook endpoint that receives the Sunshine Conversations event and creates the ticket through the API. You can write it in Python or similar, but you cannot configure it entirely inside Zendesk.

The catch: as one user noted, "there is no way to do inside of platform" for this middleware approach.

What can a messaging trigger do?

Messaging triggers are available on every Suite plan. In Admin Center, click Objects and rules in the sidebar, then select Business rules > Messaging triggers, and click Create trigger. The Run trigger dropdown has four events:

• When a customer requests a conversation
• When a message is sent
• When a conversation is added to the queue
• When a conversation is assigned from a queue

"When a customer requests a conversation" fires when the conversation is passed to an agent, so it helps once a customer asks for a human. It won't catch someone who leaves before that point. Ticket triggers are a separate tool: they run right after a ticket is created or updated, for example to notify a customer that a ticket is open.

Source: Zendesk Help Center - Creating ticket triggers and Creating messaging triggers

Tip: When building ticket triggers, keep the conditions as specific as possible. Use the "Ticket | Is | Created" condition so the trigger only runs when a ticket is created, not on every update.

How do you log every conversation with a webhook?

For teams with developers, a conversations integration gives the most control. It needs Zendesk Suite Professional or above.

How does the webhook integration work?

A conversations integration connects your own system to Sunshine Conversations. You create a webhook and choose which conversation events it subscribes to. Every matching event is sent to your endpoint, where you can log it or create a ticket. That gives you:

• A complete conversation history
• Your own data processing and storage
• A feed into external analytics tools
• Full control over how conversations are tracked

How do you set up a conversations integration?

In Admin Center, click Apps and integrations in the sidebar, then select Integrations > Conversations integrations. Click Create integration and enter a unique name. In the Webhook section, enter the endpoint URL, choose whether to include user and device data, and select the events you want to be notified about. Click Save to get the webhook ID and shared secret. Each integration can hold up to 10 API keys.

Source: Zendesk Help Center - Creating conversations integrations

This takes development work, but it is the only option here that catches a customer who leaves without clicking anything, and it can feed a data warehouse too.

Should you use a third-party AI tool instead?

If the native bot flows keep leaking conversations, a third-party AI agent that controls the whole conversation can help.

Look for a tool that offers:

• Immediate ticket creation: a ticket opens when the conversation starts
• Control over ticket rules: you decide when and how tickets are generated
• A first message: the AI greets the visitor and asks what they need, so fewer people drop off

It adds another tool to your stack, but it closes the gap where conversations die without a ticket.

What should you set up first?

Whichever fix you choose, these practices help:

1. Start with AI agent tickets

Turn on AI agent tickets first and add the Support type condition to a view. It's the fastest change and needs no code.

2. Decide which conversations become tickets

Choose between:

• All conversations as tickets (for complete analytics)
• Only escalated conversations as tickets (to keep ticket volume down)

Zendesk Chat had an option to create a ticket with the transcript attached when a chat ended. In Sunshine Conversations you add a Create Ticket action to get the same result.

Source: Zendesk Help Center - Migrating advanced AI agents

3. Create views for bot-only conversations

A dedicated view for AI agent tickets lets agents spot and follow up on them.

4. Track the drop-offs

Watch:

• The share of conversations that never get an agent response
• The bot steps where customers leave
• Satisfaction on bot-handled vs. agent-handled tickets

5. Don't flood customers with notifications

Opening a ticket for every bot interaction can send customers emails they didn't expect. Balance visibility against that by:

• Using tight trigger conditions
• Setting a sensible timeout before creating tickets
• Checking which notifications a new ticket sends

Which fix should you pick?

Turn on AI agent tickets and update your views today; that covers every conversation an AI agent handles. If customers still vanish before anything reaches an agent, the conversations integration webhook with a timeout is the only fix that catches them, and it needs Suite Professional or above plus a developer.

About Macha: Macha helps teams get more value from their Zendesk investment with tools and insights designed for modern support operations. Learn more at getmacha.com.

‍