How should managers handle an underperforming employee?

Define the performance gap before judging the person

Underperformance is a sustained or material gap between an agreed work standard and observable results or behaviour. It is not a personality label, a single disappointing day or a manager’s general feeling that someone is not proactive enough. Before the conversation, write the expected outcome, the evidence of what occurred, the impact on the work and any previous direction or support. Use dates, work products, service levels, decisions or agreed behaviours where possible. Compare like with like: a new employee, a changed role or an unexpected operating constraint may require a different interpretation from a repeated gap in familiar work. If the standard was never explicit, the manager’s first responsibility is to clarify it rather than retrospectively treat it as a failure.

• State the expected result or behaviour in terms another observer could recognise.
• Separate facts from interpretations: ‘the report arrived two days after the agreed deadline’ is evidence; ‘you do not care’ is an attribution.
• Check whether the standard is consistent with the role, available resources and how comparable work is assessed.

Diagnose the cause before choosing the response

The same performance symptom can have different causes. A missed deadline may reflect unclear priorities, a skill gap, excessive workload, a dependency the employee cannot control, avoidance of a difficult task or deliberate disregard of an agreed responsibility. Ask what the employee understood, how they approached the work, where progress broke down, what support they sought and what evidence they can share. Then test the explanation against the work. The purpose is not to excuse the gap or force a confession. It is to identify the manager action and employee action most likely to improve performance. Training helps a capability problem; it will not repair contradictory priorities, absent authority or persistent unwillingness to meet a clear standard.

• Clarity: Was the outcome, quality standard, priority, deadline and decision authority understood?
• Capability: Does the person have the knowledge, judgement or practice required for this task?
• Capacity: Are workload, tools, information, health, dependencies or operating conditions blocking delivery?
• Commitment: Is the person taking ownership, seeking help and following through on agreed actions?
• Conduct: Does the issue involve policy, safety, ethics or behaviour that needs a formal route?

Hold a direct conversation without turning it into a verdict

Meet privately and allow enough time for a real discussion. A useful opening has four parts: the purpose, the required standard, the observed evidence and the impact. For example: ‘I want to discuss the accuracy of the last three client forecasts. Our agreed standard is that assumptions are checked with the account lead before submission. In all three files, two assumptions were not verified, which led the planning team to redo the model. I want to understand what is getting in the way and agree what needs to change.’ Then pause. Ask open questions, listen for information that changes the diagnosis and summarise what you heard. Avoid a praise sandwich that hides the issue, an interrogation that presumes bad intent, or a debate about every detail. The conversation should end with shared clarity even if the employee disagrees with the manager’s assessment.

• Ask: ‘What is your understanding of the expected standard?’
• Ask: ‘Walk me through what happened and where the work became difficult.’
• Ask: ‘What will you do differently, and what support or decision do you need from me?’
• Close: ‘Here is what we have agreed, how progress will be evidenced and when we will review it.’

Build a short performance recovery plan around real work

A useful plan is specific enough to guide action and short enough to use. Define one to three priority outcomes, the behaviour or work standard for each, the actions owned by the employee, the support owned by the manager, the evidence and the review dates. Use the normal work as the practice field rather than adding generic learning tasks. If quality is the issue, review a sample before the next customer handoff. If prioritisation is the issue, require a weekly plan that makes trade-offs visible. If a skill gap is credible, combine targeted instruction with practice and feedback on an actual deliverable. Make the consequence of insufficient improvement clear without threatening or predicting an outcome outside the manager’s authority. Formal performance processes vary by organisation and jurisdiction, so align the plan with current policy and HR guidance.

• Outcome: What must be delivered, for whom, by when and to what standard?
• Employee action: What will the employee do differently in the next work cycle?
• Manager support: What clarification, feedback, access, coaching or barrier removal will the manager provide?
• Evidence: Which work product, behaviour or stakeholder signal will demonstrate progress?
• Decision point: On what date will both parties review whether to continue, adjust, recognise recovery or move to a formal next step?

Use checkpoints to improve the work, not to create surveillance

Checkpoints should match the risk and the employee’s current readiness. A high-consequence deliverable or an unfamiliar task may require an early review of approach and assumptions; familiar low-risk work may need only an agreed output review. At each checkpoint, compare evidence with the standard, ask what the employee learned and decide the next action. Do not silently take the work back, rewrite every detail or increase monitoring without explaining why. That may protect one deadline while preventing the employee from demonstrating recovery. If progress is strong, reduce oversight deliberately. If progress is weak, identify whether the plan, support, diagnosis or ownership is failing rather than extending the same arrangement indefinitely.

Example: repeated delays in a regional reporting cycle

In this illustrative fictional composite, a regional operations manager notices that a team member has submitted three monthly reports late. Instead of opening with ‘you need to be more accountable’, the manager shows the agreed submission dates and the downstream delay to finance. The employee explains that two markets now send data after the original cut-off and that she has been waiting for complete inputs. Together they separate the issues. The manager resets the escalation rule and secures earlier market owners; the employee agrees to publish an exception list by noon on cut-off day and submit the available report by 5 p.m. They review the next two cycles. If the dependencies improve but the employee still does not use the agreed escalation, the remaining gap is ownership rather than capacity.

Know when coaching is insufficient

A manager should involve HR or the appropriate formal owner when the issue is repeated after clear support, may affect employment, involves discrimination or accommodation, creates safety or regulatory risk, concerns misconduct, or cannot be handled impartially by the manager. Escalation is not a substitute for early management, and coaching is not a substitute for due process. Keep documentation factual, relevant and secure. Apply standards consistently while considering legitimate differences in role, context and required support. Singapore’s Ministry of Manpower frames fair employment around merit and performance, but this answer is manager-development guidance rather than legal advice; organisations should follow their current policies and obtain qualified advice for consequential employment decisions.

What the evidence suggests about goals and feedback

Performance management works as a continuing cycle of expectations, feedback, learning, support and accountability rather than as an isolated annual event. CIPD’s January 2026 factsheet emphasises that employees need to understand what is expected and have the motivation, skills, resources and support to succeed. Goal-setting research supports specific and appropriately challenging goals when people have commitment, feedback and the ability or strategy to perform the task. Feedback is not automatically helpful: Kluger and DeNisi’s meta-analysis found substantial variation, including cases where feedback reduced performance, and proposed that feedback is more useful when attention stays on the task rather than shifting toward threat to the self. For managers, the practical implication is to discuss observable work, enable a usable next action and review evidence—not to intensify criticism or label character.

Match the response to the likely performance barrier

Treat this as a working diagnosis to test with evidence, not a permanent label for the employee.

Common mistakes

• Waiting until an annual review to raise a gap the employee could have addressed earlier.
• Using labels such as lazy, negative or not leadership material instead of describing work and behaviour.
• Assuming a motivation problem before checking clarity, capability, capacity and dependencies.
• Giving vague advice such as be more proactive without defining the next observable action.
• Creating a plan with many goals, no manager support and no credible evidence of progress.
• Turning every checkpoint into approval-seeking or silently taking ownership back from the employee.
• Extending an ineffective informal arrangement indefinitely because the formal conversation feels uncomfortable.
• Promising a consequence, confidentiality or employment outcome outside the manager’s authority.
• Applying different standards without a role-based reason or ignoring a request that may require formal support.

Practical checklist

Frequently asked questions