Uptraining

How uptraining fits in the agent lifecycle

New agents go through classroom training, then nesting on live calls with extra support. After they move to production, three things keep performance on track: call coaching one-on-one with a team leader, quality monitoring, and uptraining for gaps that affect a group.

The usual triggers for uptraining:

• A pattern in QA results. If quality assurance shows that 40% of agents in a team skip identity verification on address changes, that is a training issue, not a coaching issue.
• A process, product or system change that existing agents were not trained on.
• A metric that drifts for a cohort, such as rising handle time or transfers among agents from the same training class.
• Compliance findings that need documented retraining.

Uptraining vs coaching vs refresher training

The terms overlap, and centers use them differently, but the practical distinction is scope. Coaching is individual and ongoing: one agent, one team leader, specific calls. Uptraining is group-based and targeted: a session for the agents who share a gap. Refresher training is often scheduled for everyone, such as an annual compliance module.

If one agent struggles, coach. If 15 agents in different teams make the same mistake, uptrain. If the whole floor makes it, look at the process, the knowledge base or the original training first.

How to run uptraining well

1. Diagnose from data. Pull the evaluations, call recordings or transactions that show the gap, and confirm it is a skill or knowledge problem rather than a system bug or an unclear policy.
2. Select by need. Train only agents who show the gap. Pulling high performers into remedial sessions wastes hours and annoys them.
3. Keep it short and practical. A typical session runs 60 to 120 minutes: a short explanation, two or three recorded call examples, then roleplay training on the exact scenario.
4. Schedule it off-peak. Uptraining time is shrinkage. Workforce management should plan it into low-volume intervals.
5. Measure the effect. Compare QA scores or the target metric for trained agents over the following two to four weeks against a baseline.

A worked example: a retail support team sees refund errors rise after a policy change. QA finds 22 of 140 agents applying the old 30-day window. A 90-minute session for those 22 agents costs 33 agent-hours. If error rates drop and each error previously took 15 minutes to fix, the session pays back quickly.

Common mistakes

Labeling uptraining as punishment makes agents defensive; frame it as a process update where possible. Running sessions without a follow-up measure makes it impossible to know if they worked. And repeated uptraining on the same topic usually points to a weak knowledge base article or job aid, not to the agents.