Understanding Why Salesforce Chats Can Be Transferred to Agents at or Over Capacity

There are different transfer options:

• Transfer to Agent
• Transfer to Skill
• Transfer to Button or Queue

All three of these options behave differently and will produce different results.

1) "Transfer to Agent" is a request from one agent to a specific agent, i.e. agent A is chatting with the customer and wants to transfer the chat to agent B.

2) "Transfer to Skill" is a request from one agent to another agent with a particular skill, i.e. agent A with the "technical support" skill is chatting with the customer and wants to transfer the chat to an agent B with the "billing specialist" skill.

3) "Transfer to Button" is a request to transfer the chat to another Chat Button, i.e. a Queue. For example, agent A is chatting with a customer and needs to transfer the customer to a different button or queue because the chat needs to go to another team. Maybe the chat started in a sales queue and now needs to go to a service queue. 

Conceptually, users can also think of chats as being "new" (i.e. a new chat that needs an agent to handle it) and "transfers" (i.e. the chat was accepted by an agent and has been transferred to another agent, through any of the methods above).

When chats are transferred directly to an agent or to a skill, these transferred chats are allowed to exceed capacity. This is done to give priority to decisions made by agents when trying to get the customer to another suitable agent. The idea is to try to ensure that the customer's interaction and issue gets resolved quickly.

For example: An agent has a configured capacity of 4 concurrent chats and they are currently handling 4 chats.
If they get a transfer request (either transferred directly to them or a skill they're a part of), they can accept the chat and "exceed" their configured capacity.

When chats are transferred to a Queue, however, they're treated like new chats, i.e. the platform looks for available agents with capacity to chat. Chats in a queue and waiting for assignment, will be routed to the next available Agent based on the Chat's age. The oldest Chat (first created) will be routed first, whether it's an initial request or a chat transfer. Back to the example above, let's say the agent has a capacity of 4 and is currently handling 4. If chats are transferred to a button or queue, the agent will not receive the chat requests (either new ones waiting to reach an agent or transfers waiting for an agent) until they have capacity.

Whether agents accept "new" or "transferred" chats, those chats count toward the agent's capacity. So any chats routed from queues will always respect the agent's capacity limit.