Capacity rules to balance the workload

A capacity rule caps how much an agent holds at once, per channel. Without one, routing keeps handing work to whoever is next.

A capacity rule caps how much an agent holds at once, per channel. Without one, routing keeps handing work to whoever is next.

Per channel, deliberately

The limits are set separately for email, messaging and calls, because they demand different things. Someone can reasonably have a dozen emails open. Two live chats is a lot. One call is one call.

Omnichannel routing considers all of them together, which is what stops an agent on a call receiving three chats.

What counts as held, on email

Only Open. A ticket on Pending, On-hold or Solved takes up none of an agent's capacity, which is correct: they are not working on it. The same statuses also make a ticket ineligible for routing in the first place. Every open email ticket counts, however it arrived and however it was assigned, including web form, API and side conversations. The ceiling is 500.

That puts real weight on status discipline. A team that leaves everything on Open looks permanently at capacity and receives nothing, and no setting will fix that.

Messaging counts differently

Here it is activity rather than status that decides. A conversation counts as inactive once the customer has not replied for ten minutes, and by default inactive conversations do not count against capacity at all. So an agent can be holding a dozen quiet conversations and still be given another.

Whether that is what you want is a decision: the setting Count inactive conversations towards an agent's capacity is off unless somebody turns it on. Off suits a team that dips in and out of slow conversations. On suits a team where a waiting conversation is still real work in somebody's head.

Calls are simpler: the only options are 0 or 1, and a call stops counting once the agent hangs up, or once wrap-up time ends if you have configured it. Setting it to 0 means that agent takes no calls.

Talk is a switch, not a number

Email and messaging each take a number up to 500. Talk takes only 0 or 1: an agent handles one call at a time or none, and 0 means they receive no calls at all. So the Talk field is a way of excluding somebody from the phone, not a way of tuning how much they take.

A call also stops counting only after the wrap-up time ends, where one is configured, rather than the moment the agent hangs up. With a long wrap-up, an agent stays at capacity after the conversation has finished, which is intended and looks like a stuck status if you do not know it.

It is not a hard ceiling

Capacity stops automatic assignment. It does not stop a person. Somebody can still be assigned work by hand beyond their limit, and an inactive conversation coming back to life can push them over it without anyone doing anything. So an agent above their number is not necessarily a broken rule.

Setting the numbers

They come from your team, not from a default. Watch what people comfortably handle, and set the limit slightly above it.

• Too low: agents idle while work waits in the queue.
• Too high: the cap does nothing and people drown.

Start conservative and raise it. Agents notice a limit that is too low and complain; a limit that is too high produces quiet burnout instead.

Different rules for different people

You add agents to a rule as assignees, so a new agent can carry less than an experienced one. That is one of the more humane things in Zendesk and it costs nothing to use.

Two mechanics to know. An agent belongs to exactly one rule, so putting somebody on a new one takes them off their old one. And there is always a default rule: new team members land on it automatically, and deleting a rule sends everybody on it back there rather than leaving them uncovered.

You will find all of this in Admin Center under Objects and rules > Omnichannel routing > Capacity rules.

See also

• Creating and managing capacity rules
• Round robin for email, messaging and calls
• Focus mode for voice and chat