Difficult Conversations, Underperformance, and Termination: A Manager's Guide

Most of the work of managing people is conversations, and the hardest of them are the ones a manager would rather not have: telling someone their work is not good enough, deciding whether to keep investing in them, and — at the end of the road — ending the relationship. This page is the practice side of that work, ordered the way it tends to unfold: why the conversation gets avoided, how to actually have it, how to give corrective feedback that lands, how to make candor a normal part of the team, when to keep coaching versus manage someone out, whether a performance improvement plan is worth running, and how to conduct a termination meeting and what to have ready for it.

A note on what kind of guidance this is. Much of the practitioner material in this area is confident and branded; the underlying evidence is uneven, and this page is honest about that section by section — some of it rests on robust, replicated research, some on credible but unproven frameworks. Where this work touches Ontario employment law — performance improvement plans and termination — the law is explained here in plain language and linked to the compliance material rather than restated or quantified. This page is general information, not legal advice.

Why do managers avoid difficult conversations and withhold bad news?

Reluctance to deliver unpleasant information is a robust, decades-old behavioural tendency known as the MUM effect, driven by fear of the recipient’s reaction, conflict avoidance, and protection of the relationship — the manager trades the employee’s long-term benefit for their own short-term comfort.

The phenomenon has been documented experimentally since 1970, when Rosen and Tesser coined the term “MUM effect”: people suppress or delay bad news even when they have no personal stake in it. In their paradigm, subjects readily relayed the neutral part of a message but avoided the unpleasant part; the effect replicated in a 1971 field study and was traced to fear of negative evaluation and discomfort at being associated with bad news. This is a genuine, replicated behavioural tendency — strong enough to treat as established — and it scales directly to managers withholding performance criticism.

The popular prevalence figures require caution. The “69 per cent of managers are uncomfortable communicating with employees” and “37 per cent are uncomfortable giving direct feedback” numbers come from a 2016 Interact survey conducted online by Harris Poll (2,058 US adults, 616 managers) and publicized by the vendor’s CEO. The release itself states the survey “is not based on a probability sample,” and Interact is a communications consultancy that sells training. These should be read as a vivid illustration of a real phenomenon, not a measured rate that transfers to Ontario small and mid-sized businesses.

In a 20-to-200-person firm the dynamic intensifies: the owner or a first-time manager often supervises people they know personally and see daily, which sharpens relationship-protection avoidance, and there is rarely an HR buffer to absorb the discomfort. Avoidance is the human default, so deliberately overriding it is itself the courageous act. What avoidance and tolerated underperformance actually cost a team is set out in the costs of avoidance and tolerated underperformance.

Source: Rosen and Tesser, On Reluctance to Communicate Undesirable Information: The MUM Effect, Sociometry 33 (1970); Tesser, Rosen and Tesser, On the Reluctance to Communicate Undesirable Messages (The MUM Effect): A Field Study, Psychological Reports 29 (1971); the survey figures from Solomon (Interact / Harris Poll), Two-Thirds of Managers Are Uncomfortable Communicating with Employees, Harvard Business Review (2016).

Confidence: the MUM effect is industry-consensus; the 69 / 37 per cent figures are illustrative only, from a single non-probability vendor survey.

Why do leaders put off tough decisions, and does delay make the problem worse?

A cluster of well-replicated decision-making biases — status-quo bias, omission bias, and choice deferral — reliably pushes leaders toward inaction on hard calls like terminations; the related claim that delay actively makes the problem worse is intuitive and consistent with those biases but is not itself precisely quantified.

Decision psychology offers a strong, replicated explanation for the deferral. Samuelson and Zeckhauser (1988) showed across experiments — and in real health- and retirement-plan choices — that people disproportionately stick with the status quo even when switching is costless and superior. Ritov and Baron (1990) isolated omission bias: people prefer harm caused by inaction over equal or lesser harm caused by action, and judge inaction as less blameworthy. Anderson’s 2003 Psychological Bulletin review synthesized four convergent decision-avoidance effects — choice deferral, status-quo bias, omission bias, and inaction inertia — into a single rational-emotional model in which anticipated regret and selection difficulty predispose people to “do nothing.” These are robust, peer-reviewed findings, warranting industry-consensus confidence for the existence of the inaction bias.

A caveat on transfer. Most of this evidence comes from lab tasks and consumer or medical choices, not from managers deciding whether to dismiss someone. The biases plausibly extend to management decisions — Anderson’s model is general — but direct experimental evidence in a real personnel context is thinner. And the second half of the question, whether delay actively makes the problem worse, is widely asserted but is not the same well-evidenced claim, so the compounding effect should be treated as directional rather than measured.

For an Ontario employer, delay also carries a concrete legal dimension: continuing to tolerate a known performance problem can undercut a later just-cause argument and lengthen reasonable-notice exposure, so inaction is not cost-free.

Source: Anderson, The Psychology of Doing Nothing: Forms of Decision Avoidance Result from Reason and Emotion, Psychological Bulletin 129(1) (2003); Samuelson and Zeckhauser, Status Quo Bias in Decision Making, Journal of Risk and Uncertainty 1 (1988); Ritov and Baron, Reluctance to vaccinate: Omission bias and ambiguity, Journal of Behavioral Decision Making 3 (1990).

Confidence: industry-consensus for the existence of inaction bias; directional for the “delay makes it worse” claim.

How do you actually conduct a difficult conversation?

The dominant practitioner frameworks converge on a small set of moves: make it safe, separate the impact of the behaviour from the other person’s intent, state the issue plainly, and inquire genuinely into the other person’s view rather than arriving certain you are right.

Two books dominate the “how to have a hard conversation” space, and both are best labelled practitioner frameworks. Difficult Conversations (Stone, Patton and Heen, 1999, from the Harvard Negotiation Project) frames every hard talk as three conversations — “What Happened?”, Feelings, and Identity — and its signature moves are to shift from certainty to a learning stance, to “disentangle intent from impact,” to abandon blame in favour of mutual contribution, and to begin from a neutral “third story.” Crucial Conversations (Patterson, Grenny, McMillan and Switzler, 2002) targets high-stakes moments and centres on “making it safe,” avoiding the “silence or violence” failure modes, and sharing your own facts before inviting the other person’s.

The common core — make it safe, separate impact from intent, state the issue plainly, inquire into the other view — is sensible and aligns with evidenced mechanisms documented elsewhere: psychological safety, which lowers the risk of speaking up, and the task-not-self focus in feedback science. Those linked mechanisms are the real warrant for the structure.

The honest gap is that there is no randomized controlled trial of either branded method against hard outcomes. For Crucial Conversations, the peer-reviewed evidence is limited to uncontrolled pre/post studies of self-reported attitudes — for example Som et al. (2021), which found statistically significant but small gains, with no control group and a 21 per cent post-response rate. For the Stone, Patton and Heen method there is no independent controlled evaluation at all. The structure is worth adopting for its plausibility and its underlying mechanisms, not because the brand is proven.

For a small Ontario firm the practical adaptations are privacy (never in a hallway), documenting specific conduct, and watching for human-rights-protected circumstances — such as disability or family status — that can surface mid-conversation. This is the operational basis for delivering a hard message “blunt but warm”: directness paired with enough care that the relationship survives.

Source: Stone, Patton and Heen (Harvard Negotiation Project), Difficult Conversations: How to Discuss What Matters Most (Viking Penguin, 1999); Patterson, Grenny, McMillan and Switzler, Crucial Conversations: Tools for Talking When Stakes Are High (McGraw-Hill, 2002); Som, Beaman, Keener and Stroup, The Impact of Crucial Conversations on Interpersonal Communication and Professionalism (2021).

Confidence: directional — credible converging practitioner frameworks, but independent controlled outcome evidence is essentially absent.

How should a manager deliver corrective feedback so it is heard and acted on?

Keep the feedback on specific, observable behaviour and its concrete impact rather than on the person’s character or worth; meta-analytic evidence shows that feedback which shifts attention to the self, rather than the task, tends to backfire.

The scientific anchor is Kluger and DeNisi’s 1996 meta-analysis (607 effect sizes; 23,663 observations): feedback improved performance on average (d = .41), “but… over 1/3 of the FIs decreased performance,” and effectiveness “decreases as attention moves up the hierarchy closer to the self and away from the task.” Feedback that lands as a judgment of the person diverts attention to ego-defence and can hurt; feedback that stays on the task helps.

The branded packaging operationalizes this principle. The Center for Creative Leadership’s SBI model — Situation, Behavior, Impact — structures feedback around when and where it happened, the specific observable behaviour, and its impact, steering managers away from personality labels. Kim Scott’s Radical Candor (2017) articulates “blunt but warm” directly as Care Personally and Challenge Directly, naming the failure modes of Obnoxious Aggression, Ruinous Empathy, and Manipulative Insincerity. Stone and Heen’s Thanks for the Feedback (2014) explains why feedback bounces off, through “truth,” “relationship,” and “identity” triggers. All are practitioner sources — sensible and consistent with the science, but frameworks rather than independently validated treatments. The thing to credit is the principle (specific, behavioural, non-character), not the brand.

A rigour check belongs here: even well-delivered multi-source feedback produces only small average improvement over time. Smither, London and Reilly (2005) found d of roughly .15 for supervisor and direct-report ratings, and near zero for self-ratings. Delivery mechanics help feedback be heard; they do not guarantee large behaviour change.

For an Ontario small or mid-sized business, behaviour-and-impact language is also the legally safer practice: documenting specific conduct rather than character judgments supports performance management and a later just-cause position and reduces human-rights exposure. The fuller science of why this works is owned by Feedback and Performance — Research and Evidence.

Source: Kluger and DeNisi, The effects of feedback interventions on performance, Psychological Bulletin 119(2) (1996); Smither, London and Reilly, Does Performance Improve Following Multisource Feedback?, Personnel Psychology 58 (2005); Center for Creative Leadership, Situation–Behavior–Impact (SBI) Feedback Model; Kim Scott, Radical Candor (2017); Stone and Heen, Thanks for the Feedback (2014).

Confidence: industry-consensus for the behaviour-not-character principle; the SBI and Radical Candor packaging is practitioner articulation, not independently proven technique.

How do you make candor normal across a team?

Routine candor is built by specific leader behaviours — visible openness, genuinely inviting input, and responding to bad news without punishing it — far more than by exhortation; it is a culture and leadership problem, not a one-conversation problem.

There is solid field evidence on which leader behaviours raise it. Detert and Burris (2007), studying 3,149 employees and 223 managers, found that managerial openness — a leader’s perceived willingness to listen to, solicit, and act on input — predicted employees’ improvement-oriented voice, and that this relationship was mediated by psychological safety. Notably, leader behaviour had its strongest effect on the voice of the best-performing employees, so a closed manager disproportionately silences the people with the most useful things to say. This is real organizational-field evidence.

Amy Edmondson’s work specifies the concrete moves that build the safety candor depends on: frame the work as a learning challenge, acknowledge your own fallibility (“I may miss something — please speak up”), invite participation with genuine questions, and respond to bad news or dissent with appreciation rather than anger. London and Smither (2002) add the systemic layer: a “feedback culture” plus individual “feedback orientation” together make feedback a normal ongoing process rather than an annual event.

For an Ontario small or mid-sized business there is both an advantage and a risk. Flat structures and daily proximity make a leader’s modelling highly visible, but one punitive reaction by the owner can freeze voice for everyone, with no other channel to fall back on. The low-cost lever is short, regular feedback rituals — recurring one-on-ones — where the leader models candor, asks for it first by soliciting feedback on themselves, and reliably rewards rather than punishes those who speak up. The goal is that hard conversations become frequent and unremarkable. The psychological-safety construct that underpins this is owned by Belonging, Psychological Safety, and the Manager-Employee Relationship.

Source: Detert and Burris, Leadership Behavior and Employee Voice: Is the Door Really Open?, Academy of Management Journal 50(4) (2007); London and Smither, Feedback orientation, feedback culture, and the longitudinal performance management process, Human Resource Management Review 12 (2002); Edmondson, The Fearless Organization (2019).

Confidence: industry-consensus.

When do you keep coaching someone, and when do you manage them out?

Coaching can fix skill, clarity, and confidence gaps — the “can’t” causes that are genuinely developable — but it does not fix a fundamental mismatch with the role or a sustained motivation problem; when honest, resourced effort over a reasonable period has not moved performance, the humane call is a clear-eyed handoff to the exit track.

The decision to keep investing in someone or to manage them out is partly evidence and largely judgment, and it helps to be honest about which is which. The coaching-effectiveness literature gives modest, real support that workplace coaching helps with skills, self-regulation, attitudes, and well-being (Theeboom et al. 2014; Jones et al. 2016), but with important caveats: many studies rely on self-report and pre/post designs vulnerable to publication bias, and the effect on hard performance outcomes is the least certain part. The takeaway is directional — coaching is worth trying for the right problems, but it is not a guaranteed fix, and the science does not tell a manager when to stop with any one person.

What coaching can fix are the “can’t” causes that are genuinely developable: a skill or knowledge gap, unclear expectations, low confidence, a fixable habit, or a workable obstacle in the environment. If the diagnosis lands here and the person is willing and trending in the right direction, keep coaching.

What coaching usually cannot fix are a fundamental mismatch between the role’s essential demands and the person’s capability (“fit”); a sustained motivation or will problem (a “won’t,” which is discipline territory, not coaching); and, critically, anything rooted in an unaccommodated disability or other protected ground, which requires accommodation rather than coaching. Beware the “set-up-to-fail” dynamic practitioner writing describes: a manager who has privately written someone off increases scrutiny, the employee becomes tentative, performance drops further, and the manager’s view is confirmed — a self-fulfilling spiral. The rule of thumb is to coach in genuine good faith or not at all.

Signals that further investment will not pay off — looking for a pattern, not one bad week — include performance that is flat or declining despite clear feedback, real support, and adequate time; the same errors recurring after being specifically addressed; the person being unable to articulate or unwilling to accept what “good” looks like even after clarification; the gap sitting in an essential, non-negotiable function of the role; and an honest “no” to the question “do I believe this person can reach standard here?” Conversely, an upward trend — even if still below target — is a strong reason to keep going.

The humane, clear-eyed handoff. When the conclusion is that coaching will not get there, dragging it out is not kindness: it prolongs uncertainty, drains the team, and, in Ontario, risks letting a positive paper record accumulate that contradicts the eventual reasons for dismissal. Two honest reframes help — some people are simply in the wrong role or the wrong company, and saying so plainly is more respectful than a year of ambiguous feedback; and the decision to stop is a judgment call to own, not to dress up as data. Before the handoff, confirm three things: that the accommodation question was properly considered and is not the real issue; that the gap is “can’t / fit,” not “won’t” (which would route to discipline instead); and that the documentation honestly reflects the support given and the lack of improvement. Then move to the exit track, where the legal mechanics of notice, severance, and a respectful termination live — they should not be improvised.

For a Kitchener-Waterloo small or mid-sized business, one sustained underperformer is disproportionately visible and costly, and the temptation is either to avoid the conversation for months or to lurch straight to firing. Both are mistakes. The better path is a real, time-bound coaching effort with honest feedback; a decision at the end of it; and, if the answer is “manage out,” doing it promptly, respectfully, and with proper notice and severance and legal advice — a clean, well-documented, good-faith exit is cheaper and kinder than a drawn-out one.

This work connects outward: the diagnosis of underperformance and how to manage it, the meta-analytic evidence on workplace coaching, and the duty to accommodate disability that must be ruled out before any “manage out” decision.

Source: Theeboom, Beersma and van Vianen, Does coaching work? A meta-analysis on the effects of coaching on individual level outcomes in an organizational context, The Journal of Positive Psychology 9(1) (2014); Jones, Woods and Guillaume, The effectiveness of workplace coaching: A meta-analysis of learning and performance outcomes from coaching, Journal of Occupational and Organizational Psychology 89(2) (2016); Mager and Pipe, Analyzing Performance Problems (3rd ed., 1997).

Confidence: directional — modest empirical support for coaching, with the “when to stop with one person” call being judgment, not data.

Do performance improvement plans (PIPs) actually work, and how do you run one in good faith?

There is very little rigorous, peer-reviewed evidence that PIPs rehabilitate performance; in practice they frequently function as a documentation device and a runway to exit, and employees often read them that way, so a manager should be honest with themselves about which kind of plan they are really running.

What exists is overwhelmingly practitioner, HR-vendor, and legal commentary — not controlled studies showing that putting people on PIPs makes them good at their jobs. That is thin evidence, not “mixed” evidence; the volume of confident HR content should not be mistaken for the question having been answered. Any claim that “PIPs work” is best treated as an assertion of practice, not a finding.

In practice the PIP is frequently a documentation device and a runway to exit, and this is said plainly by sources on every side. Practitioner and legal commentary repeatedly describes PIPs as “commonly understood by many employees to be the first step to termination.” Even SHRM’s own expert acknowledges the framing: at SHRM25, James McDonald (SHRM-SCP) cited the Wall Street Journal’s November 2024 article by Lauren Weber and Chip Cutter, “The Most Hated Way of Firing Someone Is More Popular Than Ever. It’s the Age of the PIP.” Ontario employment lawyers are blunter, noting that some employers use a PIP “as a cover for their intent to terminate” — to manufacture cause or pressure a resignation. The honest framing is therefore that a PIP is mostly a legal and procedural instrument — it documents that the employee was warned and given a chance — and only sometimes a genuine improvement instrument. A manager should be clear with themselves which one they are running.

If a PIP is to be used, it should be the good-faith version. Drawing the common threads from SHRM, employment-law commentary, and general practitioner guidance, a defensible PIP is:

1. Specific and measurable — observable behaviours and outcomes, not “improve your attitude,” with targets tied to the real standard of the role.
2. Genuinely achievable — goals a reasonable person in that role could hit in the time given. Unrealistic or arbitrary targets are the hallmark of a sham.
3. Time-bound but reasonable — SHRM’s James McDonald advises a minimum of 30 days, “but 60 or 90 days is more customary,” with the length fitting the role and the goals (“Don’t put someone on a three-year PIP” — and, equally, a one-week deadline for meaningful change reads as a setup).
4. Resourced — support, not just a deadline — training, mentoring, clarified expectations, and regular check-ins. A PIP with consequences but no support is the single biggest tell that it is only a paper trail.
5. Co-created and documented — involve the employee, record progress honestly (improved / same / worse), and do not belittle or ignore obstacles they raise.
6. Diagnosed first — only for genuine “can’t” performance gaps where improvement is plausible. A PIP is the wrong tool for misconduct (which routes to discipline) and the wrong tool where the real cause is an unaccommodated disability (which routes to accommodation first).

A sham PIP creates real legal exposure in Ontario. A PIP built to fail — arbitrary standards, no support, a surprise rollout, unrealistic timelines, or one timed to deflect an employee’s own complaint — can support claims for constructive dismissal and bad-faith or moral damages. In Piresferreira v. Ayotte, 2010 ONCA 384, the Court of Appeal for Ontario upheld damages for mental suffering (assessed under the Honda v. Keays manner-of-dismissal framework) where a supervisor at Bell Mobility abused an employee and then strategically timed a negative PIP to thwart her complaints; the employer’s handling conveyed that “its priority was the PIP, not [the supervisor’s] behaviour” (leave to appeal to the Supreme Court of Canada was refused). In Brake v. PJ-M2R Restaurant Inc., 2016 ONSC 1795, aff’d 2017 ONCA 402, a long-service McDonald’s manager was held to have been “set up to fail” on a PIP with effectively unattainable standards; the court found no cause and awarded a long notice period.

Separately, Ontario law sets a high bar to dismiss for cause on performance grounds: a single bad review or failed PIP is almost never enough, and even a fair, failed PIP generally does not eliminate the employee’s entitlement to notice or severance — for-cause performance terminations require wilful misconduct under the Employment Standards Act, which incompetence rarely meets. The statutory notice and severance numbers, the just-cause mechanics, and the documentation that holds up are governing material elsewhere: see Ontario Employment Standards: Termination and Severance, when misconduct becomes just cause, progressive discipline and documentation, disciplinary documentation that holds up, and the duty to accommodate disability.

For an Ontario small or mid-sized business: if you put someone on a PIP, decide honestly whether you believe they can succeed. If yes, resource it like you mean it and you may keep a good employee. If no, recognize that you are really building a termination record — and that an unfair or bad-faith PIP can increase what you owe (moral or aggravated damages on top of notice), not protect you. Either way, assess the accommodation question first, keep contemporaneous documentation, and get legal advice before terminating on the strength of a failed PIP. This is general information, not legal advice.

Source: SHRM, 8 Steps for Effective Performance Improvement Plans (James McDonald, SHRM-SCP, 2025); Piresferreira v. Ayotte, 2010 ONCA 384; Brake v. PJ-M2R Restaurant Inc., 2016 ONSC 1795, aff’d 2017 ONCA 402 (McMillan LLP case comment); Achkar Law (Ontario), Performance Improvement Plans: Rights and Risks (2023).

Confidence: directional — rehabilitation evidence is thin; the case-law and legal-exposure points are well-established Ontario authority.

How do you conduct a termination meeting?

By the time you are in the room the decision is made and final — the meeting exists to deliver it, not to re-open it. Get to the point within the first 30 to 60 seconds, keep the meeting to roughly 10 to 15 minutes, do not debate the decision, and treat the person with visible respect on the way out.

Everything that makes a termination go badly traces back to forgetting that the decision is final. The decision, the entitlements paperwork, and the logistics should be prepared in advance so the meeting itself can be short, clear, and humane.

Get to the point in the first minute. Practitioner guidance is strikingly consistent: deliver the news within the first 30 to 60 seconds, before any small talk creates false hope. SHRM’s Paul Falcone and Canadian counsel alike advise stating the decision plainly — “we’ve made the decision to end your employment, effective today” — rather than burying it under preamble. Brené Brown’s maxim, quoted widely in this literature, captures the principle: “Clear is kind. Unclear is unkind.” A garbled, over-cushioned message leaves the person confused and more upset, not less.

Keep it brief and do not debate the decision. Most HR sources put the meeting at roughly 10 to 15 minutes. The decision is not up for negotiation in the room; if the person argues, acknowledge their feelings and restate that the decision is final. Language like “you will be terminated,” which implies the outcome is still pending, is best avoided.

“Blunt but warm” is a balance, not a contradiction. Blunt means the decision is stated clearly and treated as final. Warm means the delivery is respectful, calm, and unhurried in tone — the person is allowed to react, is listened to, and is not stripped of dignity on the way out. Falcone frames the goal as letting the individual “retain dignity and self-respect.” HBR’s Joel Peterson, chairman of JetBlue, stresses that this conversation should not be outsourced to HR and should be handled by the person’s own manager. The two are not in tension: a manager can be unambiguous about the outcome and decent about the manner at the same time.

Without-cause versus for-cause changes the script, not the humanity. In Ontario the large majority of dismissals are without cause — the employer alleges no wrongdoing and simply provides the required entitlements. In a without-cause meeting, do not improvise, over-explain, or invent a performance narrative; a brief, true framing (“this is a business decision, not a reflection on any single failing”) is enough, and saying more can create legal exposure. A for-cause termination is different and higher-risk: it rests on the discipline and investigation record built beforehand, and the reason is stated by reference to that documented record. Either way, the meeting is conducted with the same respect.

Have the package and logistics ready. The entitlements, final-pay information, benefits continuation, and any separation documents should be printed and ready to hand over, with digital copies to follow. The legal substance of those entitlements — ESA notice and termination pay, statutory severance, common-law reasonable notice, and the enforceability of any termination clause — is governing law and lives in Ontario Employment Standards: Termination and Severance; it should be cited and linked, not restated or quantified here. Why the manner of dismissal itself carries legal weight is covered in dignity and risk in termination.

For an Ontario small or mid-sized business, the practical discipline is preparation plus restraint: decide, document, prepare the package, pick a witness, and then in the room say less rather than more. The single most common own-goal is a manager who, trying to be kind, improvises reasons or makes unplanned promises that later become evidence. This is general information, not legal advice.

Source: SHRM, How to Have the Termination Discussion (Paul Falcone, 2021); Harvard Business Review, Firing with Compassion (Joel Peterson, 2020); Rubin Thomlinson LLP, Employee Dismissals & Terminations: Up in the Air (Christine Thomlinson, 2014).

Confidence: industry-consensus.

The termination meeting checklist: what to have ready and what not to do

Before the meeting, have the entitlements, paperwork, a witness, the logistics, and a communication plan ready; in the room, avoid the recurring traps of over-explaining, making unplanned promises, arguing the decision, and careless timing. A termination goes wrong less from what is said than from what was not prepared.

The checklist below is practitioner and professional-body consensus, not a legal code; where an item is a genuine legal risk it is flagged and linked to the compliance material.

Before the meeting — have ready:

• The documentation. The decision record, and (for cause) the progressive-discipline trail. The HRPA terminations guideline expressly recommends that HR professionals consult legal counsel in advance of and when anticipating a termination.
• The final-pay and entitlements package, printed and ready to hand over, with digital copies to the person’s personal email. The HRPA guideline flags as a failure case a meeting where the employee is left unclear on termination pay, when benefits expire, and the related RRSP and benefit calculations. The amounts and clause enforceability are compliance material — link, do not restate.
• A witness. A second management or HR representative should attend, both to support the conversation and to provide a credible account if events are later disputed. Weigh this against the “ganging up” perception in borderline cases, but the default is to have one present.
• The room and timing. Private, out of sight and earshot of the team. Canadian practitioners advise against Friday afternoons and the eve of a holiday — a person let go on Friday has the whole weekend to brood with no access to a lawyer, advisor, or counsellor. A Tuesday morning is the commonly recommended alternative.
• The logistics of departure: return of property (laptop, keys, cards, phone), system and access deactivation timing, and how the person will collect personal belongings.
• A communication plan for the team and clients: a brief, general, pre-agreed statement about who will cover the work, protecting the departing person’s privacy.

In the room — the recurring traps:

• Over-explaining. In a without-cause termination, do not improvise reasons or build a performance narrative on the spot. This is both a dignity issue and a legal risk: loose or shifting explanations can be used against the employer.
• Unplanned promises. Do not make commitments about references, payouts, or future work that were not decided in advance. Casual, well-meaning comments about benefits have led to claims; Wolters Kluwer’s guidance specifically warns that more than a few employers have been sued for benefits they did not intend to provide, on the strength of an offhand remark.
• Arguing the decision. Acknowledge emotion, listen, but restate that the decision is final. The meeting is about employment status and next steps, not re-litigation.
• Careless timing and the “perp walk.” Avoid marching the person out under escort except where genuinely warranted (for example, theft or a safety concern). Let them leave with dignity and arrange property return discreetly.
• Doing it by text, email, or — if avoidable — video. Face-to-face is the consensus standard of respect.

After the meeting: deactivate access on the agreed timeline, issue the Record of Employment to Service Canada within five days of the last day, send the final pay by the next pay date, update records, and execute the team and client communication plan.

For an Ontario small or mid-sized business without a dedicated HR department, the highest-value move is a one-page pre-termination checklist plus a brief legal review of the plan before the meeting — the cost of that review is trivial next to a wrongful-dismissal or bad-faith claim. This is general information, not legal advice.

Source: HRPA Practice Guideline, Terminations Checklist (2022); SHRM, Termination Best Practices: A Toolkit to Protect Your Business (2023); Wolters Kluwer, Conducting a Termination Meeting (2022).

Confidence: industry-consensus.

---

This page is general information about managing the people side of an Ontario business, not legal advice. The sections on performance improvement plans and termination touch governing employment law; the legal substance — ESA notice and termination pay, statutory severance, common-law reasonable notice, just cause, and the duty to accommodate — lives in the linked compliance material, and the area is one where the gap between the statutory minimum and what a court may order is wide, so obtain advice before acting on any of the above.