Zendesk Guided Mode

Normally an agent opens a view, scans the list and picks. Guided mode removes the list. The agent starts play mode on a view and Zendesk serves tickets one at a time in the order the view sorts them, with limited ability to browse around or skip ahead.

It is configured on the agent role rather than per person, and custom agent roles are a higher-tier feature, so check what your plan includes before planning a rollout. The precise restrictions have varied across releases too, particularly around search and what an agent can reach outside the served ticket, so verify the current behaviour in the Zendesk documentation rather than assuming.

The mechanism is simple. The consequences are not.

Experienced agents batch. They spot four tickets about the same outage and answer all four in eight minutes, because the thinking happens once. Guided mode serves those four an hour apart, interleaved with unrelated work, and the same job takes twenty-five minutes.

It also punishes mixed queues. If one ticket needs thirty seconds and the next needs an hour of investigation, forcing strict order means the quick wins pile up behind the hard one, and both your agents and your response times suffer.

And it can feel like distrust. That reaction is not irrational, because the feature does encode a judgement about whether agents can be relied on to pick well. Introduce it without explaining why and you'll spend the next fortnight managing morale instead of tickets.

The genuinely hard case is the duplicate. Two tickets about the same problem, served an hour apart to two different agents, and neither can see the other. In a browsable list somebody might have spotted the pair. In guided mode, nobody is looking at the list.

Start with one role and one team, not the whole floor. Tell people why, in terms of the problem rather than the tool. Agree a review date up front, because a trial with no end date is just a change.

Watch two numbers: resolution time on your oldest tickets, which should improve, and tickets solved per agent per hour, which may dip. If the first improves and the second holds, it is working. If both get worse, your queue is more varied than you thought.

And keep an escape hatch. There will be a Monday morning outage when every agent needs to see the whole queue at once, and a role you cannot switch off in five minutes will make that morning considerably worse.

One more thing to prepare: the tickets an agent genuinely cannot handle. In a browsable queue they skip it and somebody else picks it up. In guided mode they need a documented route out, usually a macro that reassigns to the right group with an internal note. Without that route, agents either sit stuck or start guessing, and guessing on a ticket you don't understand is how bad answers happen.