How to Stop Agents Cherry-Picking the Easy Tickets

To stop agents cherry-picking the easy tickets, take the choosing away from a shared queue and let rules assign work instead: route by skill and current workload, and make every assignment visible. When tickets are assigned rather than grabbed, the easy-hard imbalance goes away because no one is choosing.

What cherry-picking looks like

Cherry-picking happens when agents work from an open queue they can pull from freely. Faced with a mix of quick wins and thorny cases, people reach for the quick wins. Their handle time looks great, their ticket count is high, and the hard tickets, the angry ones, the ones needing deep product knowledge, drift to the bottom and age. A few conscientious agents quietly absorb the tough work while others coast on volume. Morale and SLAs both suffer, even though everyone is technically busy. That uneven difficulty load is one of the clearest paths to agent burnout, and headcount metrics will not show it because the ticket counts look fine.

Why it is a system problem, not a people problem

It is tempting to frame cherry-picking as an attitude issue, but the shared open queue practically invites it. If you give people a choice and reward raw ticket count, they will optimize for easy tickets. The durable fix is not lectures or leaderboards; it is removing the choice at the point of assignment. When a ticket is assigned to a specific agent by a rule, there is nothing to cherry-pick.

Move from "pull" to "assign"

The core change is to switch from a pull model (agents pick from a queue) to an assign model (a rule routes each ticket to an agent). This is the heart of support ticket routing. Instead of a shared pile, each incoming ticket is evaluated and sent to one eligible agent based on rules you set. The easy tickets and the hard tickets are distributed by the same logic, so no one can skim the top. For the step-by-step mechanics, see how to auto-assign help desk tickets.

Route by skill so hard tickets land well

Assigning tickets blindly would just spread the hard ones randomly, which is not much better. The point is to send each ticket to someone who can handle it. Skills-based routing tags agents with what they know, billing, integrations, tier-two technical, and routes each ticket to an eligible agent. The hard tickets go to people equipped for them, not to whoever happens to be least likely to dodge.

Balance by workload, not by ease

When agents choose, they self-balance toward easy work. When rules assign, you balance deliberately by current load. Workload balancing looks at how many open tickets each eligible agent already holds and favors those with capacity, so a fast agent who cleared their queue gets the next ticket rather than the agent stuck on a complex case. This measures real load, not the misleading signal of raw closed count.

Make it visible so it feels fair

Agents accept assignment far more readily when they can see the logic. Transparent assignment shows why each ticket landed with each person, matched on skill, chosen by lowest open workload. That transparency does two things. It removes the suspicion that assignment is arbitrary, and it lets you defend the system when someone complains they always get the hard ones. If that complaint is true, the record will show it and you can adjust the rules.

Watch for the failure modes

• Gaming the status. If agents mark themselves unavailable to dodge assignment, pair availability with light oversight so status reflects reality.
• Reopen churn. Cherry-pickers sometimes close fast and reopen later. Track reopens alongside closes so speed is not the only metric that counts.
• Skill gaps. If only one agent can handle a category, assignment will overload them. Use it as a signal to cross-train.
• Fallback needs. When no eligible agent is free, route to a supervisor queue rather than dropping the ticket.

Measure the right things

Ditch raw ticket count as the headline metric; it is exactly what rewards cherry-picking. Instead watch the spread of ticket difficulty across agents, aging on the hardest categories, and first-response time on complex tickets. If assignment is working, the hard tickets stop piling up and the load looks even across the team. Industry support-ops commentary suggests that rules-based assignment tends to reduce the easy-hard imbalance compared with open queues, though the effect varies by team, so verify with your own data rather than assuming a number.

Assigner replaces the grab-what-you-like queue with rules you control, routing each ticket to an eligible, available agent by skill and current workload, and recording why. The AI assists with matching; the rules stay yours and nothing is guaranteed. To see it in a support context, explore support ticket routing or review the pricing page. If you are weighing whether an AI layer would help here, our breakdown of what AI ticket routing actually does covers where classification helps and where it changes nothing. The same shared-queue problem shows up in staffing, where reqs get claimed instead of assigned: assigning job orders to recruiters works through the desk-ownership version of it.