Does Practicing a Difficult Conversation Before You Have It Work? What the Research Says

This article is part of our guide to conflict practice: practicing a difficult conversation before you have it.

Short answer: does practicing a difficult conversation work?

Yes. In the study that tested practice against a simulated counterpart directly, people who practiced used escalating tactics about two-thirds less and cooperative tactics about twice as often in a live conflict afterwards, compared with people who had the same training but no practice. Around that study sits a large body of evidence that practicing a skill by doing it, with someone playing the other side and feedback afterwards, beats reading or watching about it, in medicine, negotiation and assertiveness training.

Two things make it work: a counterpart that pushes back, and a verdict afterwards. Practice without feedback can make some things worse, which is why the verdict matters. You decide what you stand for; practice makes sure you can hold it when it counts.

The study that tested it directly

In 2024, a Stanford team (Omar Shaikh, Valentino Chai, Michele Gelfand, Diyi Yang and Michael Bernstein) published a study at CHI, the main human-computer interaction conference, called Rehearsal: Simulating Conflict to Teach Conflict Resolution. It is the closest thing to a direct test of the question in this article's title.

The design. Forty adults, recruited online, were split into two groups. Everyone watched a four-minute training video on a standard conflict framework (interests, rights and power) and got a written summary of the strategies. Half of them, in addition, spent up to twenty-five minutes practicing three conflict scenarios against a language-model counterpart built to play the other side. Then everyone had a live, ten-turn text conversation about a new conflict with one of the researchers, who played the counterpart without knowing which group each person was in and followed a fixed script: open contentious, retaliate against contentious moves, give ground when cooperative moves were repeated. Participants were told they were talking to a real person.

The result. Across the live conversation, the group that had practiced reduced its use of escalating, competitive strategies by an average of 67% and roughly doubled its use of cooperative ones. The two cooperative strategies the training taught, stating interests and making proposals, were used about twice as often (around 3 times per conversation against 1.5). Power moves ran at about a third of the control group's rate.

Two findings that matter as much as the headline. First, the practice group did not know more: on a written quiz about the framework, the two groups scored about the same. Practice built what the authors call street knowledge rather than book knowledge; it changed what people did, not what they could recite. Second, everyone's confidence dropped after the live conflict, in both groups. The authors call it a conflict reality check. Practice did not make people feel more confident; it made them behave better while feeling about the same.

What decades of role-play research show

The Stanford study is new because the counterpart was a model. The idea that practicing a conversation against someone playing the other side beats reading about it is not new at all; it is how medicine, nursing and negotiation have taught hard conversations for decades, and that literature is large.

Medical and nursing training. A 2015 meta-analysis of eighteen controlled trials (1,326 nursing students) found that practicing with standardised patients, trained actors who play the patient, improved communication skills by a very large margin (an effect size of 1.86, where 0.8 is conventionally called large), with smaller but real gains in confidence and clinical competence. Most of the trials were not randomised, which is the usual weakness of this field. A 2011 overview of twelve systematic reviews on teaching communication skills to physicians reached a blunter conclusion: programmes worked when they focused on practicing skills, and the best strategies within them were role-play, feedback and small-group discussion; presentations, demonstrations and written material should only be used in support.

Does the counterpart have to be a professional? A 2025 meta-analysis compared practicing with trained actors against practicing with a fellow student playing the role, across ten studies and 721 learners. On most outcomes there was no meaningful difference; the actors only won on learners' self-confidence. That is a useful result for anyone practicing a difficult conversation at home: the counterpart does not need to be good at acting. It needs to hold the other side.

Negotiation. A 2010 review of six meta-analyses covering fifty-seven laboratory experiments found that the amount of negotiation training strongly predicted individual performance (a correlation of 0.76 for long training against 0.22 for short), and most of that training is role-play. Negotiation is the field that has taken practice most seriously, which is one reason negotiators walk into hard conversations better prepared than most couples do.

Public speaking. The same logic holds for the fear itself. A 2022 meta-analysis of eleven studies found that practicing in front of a virtual audience reduced public-speaking anxiety about as much as practicing in front of a real one, and both worked well. Exposure to the feared situation, in a form where nothing is at stake, is the active ingredient.

Planning the moment: implementation intentions

A separate line of research explains part of why practice works even when it is only in your head. Implementation intentions are if-then plans: if he says it is not the right time, then I will say I understand and that I would still like to set a time. A 2006 meta-analysis of ninety-four tests found a medium-to-large effect of such plans on people achieving what they set out to do (an effect size of 0.65). A 2025 follow-up meta-analysis across 642 tests found effects ranging from small to large across different outcomes, and identified what made them bigger: an explicit if-then format, caring about the goal, and, directly relevant here, having practiced the plan.

Conflict practice is implementation intentions with a counterpart. You are not only deciding what to say; you are pairing each thing the other person is likely to say with what you will do when they say it, and running the pair until it is automatic. That is the mechanism behind step two of the seven-step method: pick the real objection, not the convenient one.

Assertiveness training, the older clinical tradition that this practice descends from, used behavioural practice as its core method: say the line, get the response, try again. A 2026 meta-analysis of twelve randomised trials found a medium effect of assertiveness training on social anxiety. A 2018 review argued that assertiveness training is a forgotten evidence-based treatment whose research dried up in the 1980s, not because it stopped working but because the field moved on.

Practice without a verdict can make things worse

This is the finding worth pausing on, because most people who practice a conversation do it without any standard applied afterwards.

In a 2026 randomised study at CHI, ninety-four novice counselors practiced with a language-model client. Half of them also got structured feedback after each practice session; half just practiced. The feedback group improved on two core skills. The practice-only group did not improve on either, and on one measure, empathy, they got worse: a medium-sized decline. Practicing a conversation with no verdict did not just fail to help; it drifted the skill in the wrong direction.

A second 2026 study, with 145 managers practicing difficult workplace conversations against a model built to push back, found that confidence rose and avoidance fell after practice, and that the full system with feedback beat practice alone on a new scenario. The authors also measured avoidance in a way worth remembering: about six in ten managers said they would do extra administrative work to avoid a difficult conversation, and about half would give up money.

The lesson for anyone practicing at home: the counterpart matters, and the verdict matters more. Practice tells you what you did. A fixed standard tells you whether it was any good, and the exact sentence where it stopped being good. That is why AET01 Conflict scores every run against the same standard, and why the one thing an assistant cannot do for you is apply one.

AET01 Conflict

Fix the fight before you have it.

Practice it against an AI partner who doesn't back down. Two minutes, a score, and the line that would have held.

Open AET01 ↗Coming to iOS and Android

What avoiding the conversation costs

The research above is about having the conversation better. The reason to have it at all is better documented by surveys, and it is worth knowing that these are vendor surveys rather than peer-reviewed studies, so treat the numbers as indicative.

A 2016 survey of 1,025 people by the company behind Crucial Conversations found that respondents estimated an avoided conversation cost their organisation about $7,500 on average, and that half of them wasted seven or more working days on it; four in ten said they stewed on it for two weeks or more. The same company's 2022 survey of 1,100 people found the conversations most avoided were about a colleague not pulling their weight (68%) and underperformance (66%), and that only 9% rated themselves or their colleagues very skilled at having them. A 2019 survey of more than five hundred US employees found seven in ten were avoiding a difficult conversation at work; those who stayed silent were far more likely to complain to colleagues or post a negative review instead.

The relationship research says the same thing with better methods. A twelve-month study of sixty-one couples found that direct strategies for asking a partner to change felt unsuccessful in the moment but predicted more actual change over the year, while soft, indirect approaches felt successful at the time and produced none. A two-week daily-diary study found that giving in to avoid conflict or a partner's displeasure was linked to lower well-being and relationship quality, and was particularly damaging over time, while giving for positive reasons was not. In plain terms: the quiet trade that ends the argument costs more than the argument.

What this means for the conversation with your partner

Most of the direct evidence is about work: counselors, managers, negotiators, nurses. The relationship research is about the patterns, not about practice. So what carries over?

The pattern research is unambiguous that how couples have the conversation matters. Across seventy-four studies and more than fourteen thousand people, the demand-withdraw pattern, where one partner pushes for change and the other goes quiet, is moderately and consistently linked to worse relationships, and it is the role, not the gender, that does the damage. The longest follow-up of a communication-skills programme for couples, eleven years, found that couples who had taken five sessions of communication and conflict-management training separated or divorced at about half the rate of those who had not. A 2019 meta-analysis of thirty-three randomised trials found couples therapy improves relationship satisfaction by a medium effect, with emotionally focused work holding its gains better over time.

What that suggests, without over-claiming: the skills that practice builds, saying the need without the accusation, staying in the conversation past the point where you would normally withdraw or cave, are the same skills the couples research says matter. Practice is a way to build them between sessions. It is not a substitute for the work on the pattern underneath, which is what couples therapy is for. How to practice the argument with your partner →

What the research does not show

• That practice fixes the relationship. It changes how you hold a position in one conversation. The cycle that produces the same fight every week is a different problem, with its own evidence base.
• That the effect lasts. The direct study measured one conversation immediately after practice. The medical literature suggests practiced skills persist, but nobody has followed people who practiced a conflict for a year.
• That a text chat is a kitchen. Face-to-face conflict has a body in it. Practice over text or voice is a simulator; it is closer to the real thing than reading, and further from it than the real thing.
• That the position is right. Practice makes you better at holding a line. Whether the line is the right one is a question for you, and sometimes for the person on the other side of it.
• That a model's feedback is correct. The counselor study is the warning: feedback helped because it was structured against a standard. A model that praises you is not feedback.

Nothing here is therapy, and practice is not for crisis.

FAQ: practicing difficult conversations

Does practicing a difficult conversation actually help?

Yes. In the study that tested practice against a simulated counterpart directly, everyone got the same short training and half also practiced; in a live chat conflict afterwards the practice group used escalating tactics about two-thirds less and cooperative tactics about twice as often. Role-play with feedback has decades of supporting evidence in medical, nursing and negotiation training.

What did the Stanford conflict study find?

That practicing conflict scenarios against a simulated counterpart changed behaviour without changing knowledge. The practice group used fewer competitive strategies and about twice as many cooperative ones in a live conflict, scored no better on a written quiz, and felt no more confident afterwards. It was a forty-person, single-session study using text chat, published at CHI 2024.

Is practicing with a friend as good as practicing with a professional?

Probably, for most outcomes. A 2025 meta-analysis across ten studies found no meaningful difference on skills between practicing with trained actors and practicing with a peer playing the role; the actors only improved learners' self-confidence. What matters is that the counterpart holds the other side and that you get a verdict afterwards.

Can practicing a conversation make things worse?

It can, if there is no feedback. In a 2026 randomised study of ninety-four novice counselors, those who practiced with a simulated client but got no structured feedback did not improve and got measurably worse at empathy, while those who got feedback improved. Practice needs a standard applied afterwards, not just repetition.

Why do I rehearse arguments in my head, and does that count as practice?

The planning research says the useful half of it helps: if-then plans for what you will do when the other person says X have a medium-to-large effect on follow-through, and the effect is larger when the plan has been practiced. The rest is covered in its own article: why you rehearse arguments in your head →

How many times should I practice a hard conversation?

The direct study used up to twenty-five minutes across three scenarios, once. The training research consistently finds that more practice helps and that short, repeated runs with feedback beat one long one. A practical rule: two minutes at a time, a verdict, and three or four runs across a week before the real conversation.

Does practice help with the conversation with my partner specifically?

The direct evidence is from work settings; the couples research is about patterns rather than practice. But the skills are the same ones the couples research says matter: saying the need without the accusation and staying in the conversation instead of demanding or withdrawing. The demand-withdraw pattern is consistently linked to worse relationships across seventy-four studies, and communication-skills training for couples has the longest positive follow-up in the field, eleven years. Practice is a way to build the skill between sessions, not a replacement for working on the pattern.

Next step

If you have a conversation you have been drafting and not having, the research says: practice it against something that pushes back, get a verdict, and go again. AET01 Conflict is an app for practicing difficult conversations, built for exactly that, for partners and co-founders: a counterpart that holds its position, two minutes, a score against a fixed standard and the line that would have held. What conflict practice is → · Why general AI fails as a practice partner →

If the conversation is the same one you have had for years, the pattern is the problem, and that is what couples therapy and co-founder coaching are for.

Research and further reading

• Shaikh, O., Chai, V., Gelfand, M. J., Yang, D. & Bernstein, M. S. (2024). Rehearsal: Simulating conflict to teach conflict resolution. Proceedings of the CHI Conference on Human Factors in Computing Systems (CHI '24). https://doi.org/10.1145/3613904.3642159. Forty participants; practice against a simulated counterpart reduced competitive strategies by 67% and doubled cooperative ones in a live chat conflict, with no difference in quiz knowledge. Forty participants, one session, a live text-chat conflict with a scripted counterpart.
• Louie, R., Shah, R. S., Orney, I. H., Pacheco, J. P., Brunskill, E. & Yang, D. (2026). Can LLM-simulated practice and feedback upskill human counselors? A randomized study with 90+ novice counselors. CHI '26. https://doi.org/10.1145/3772318.3791821. Practice plus feedback improved skills; practice alone did not, and empathy declined.
• Kimani Cross, E., Santos, L., Denoue, L., Carter, S., Bravo, N. S. & Sieck, K. (2026). ConvoDojo: Structured LLM-based sparring partners for difficult workplace conversations. ACM Conversational User Interfaces (CUI '26). https://doi.org/10.1145/3816046.3816230. 145 managers; confidence rose and avoidance fell after practice; the full system with feedback beat practice alone on a new scenario.
• Oh, P.-J., Jeon, K. D. & Koh, M. S. (2015). The effects of simulation-based learning using standardized patients in nursing students: A meta-analysis. Nurse Education Today, 35(5). https://doi.org/10.1016/j.nedt.2015.01.019. Eighteen trials, 1,326 students; communication skills d = 1.86.
• Berkhof, M., van Rijssen, H. J., Schellart, A. J. M., Anema, J. R. & van der Beek, A. J. (2011). Effective training strategies for teaching communication skills to physicians: An overview of systematic reviews. Patient Education and Counseling, 84(2). https://doi.org/10.1016/j.pec.2010.06.010. Role-play, feedback and small-group discussion were the effective strategies; written material only in support.
• Xiao, J. & Fu, X. (2025). Is the use of standardized patients more effective than role-playing in medical education? A meta-analysis. Frontiers in Medicine, 12, 1601116. https://doi.org/10.3389/fmed.2025.1601116. Ten studies, 721 learners; no meaningful difference on skills between actors and peers.
• ElShenawy, E. (2010). Does negotiation training improve negotiators' performance? Journal of European Industrial Training, 34(3), 192–210. https://doi.org/10.1108/03090591011031719. Six meta-analyses over 57 experiments; long training correlated 0.76 with individual performance.
• Reeves, R., Curran, D., Gleeson, A. & Hanna, D. (2022). A meta-analysis of the efficacy of virtual reality and in vivo exposure therapy as psychological interventions for public speaking anxiety. Behavior Modification, 46(4). https://doi.org/10.1177/0145445521991102. Eleven studies; virtual and real audiences reduced anxiety about equally.
• Gollwitzer, P. M. & Sheeran, P. (2006). Implementation intentions and goal achievement: A meta-analysis of effects and processes. Advances in Experimental Social Psychology, 38, 69–119. https://doi.org/10.1016/S0065-2601(06)38002-1. Ninety-four tests; d = 0.65.
• Sheeran, P., Listrom, O. & Gollwitzer, P. M. (2025). The when and how of planning: Meta-analysis of the scope and components of implementation intentions in 642 tests. European Review of Social Psychology, 36(1), 162–194. https://doi.org/10.1080/10463283.2024.2334563. Effects of d = 0.27 to 0.66; larger when plans were rehearsed.
• Hede, Malouff & Meynadier (2026). A meta-analysis of randomised controlled trials on the efficacy of assertiveness training for social anxiety. International Journal of Applied Positive Psychology. https://doi.org/10.1007/s41042-026-00297-7. Twelve trials, 503 participants; g = 0.62.
• Speed, B. C., Goldstein, B. L. & Goldfried, M. R. (2018). Assertiveness training: A forgotten evidence-based treatment. Clinical Psychology: Science and Practice, 25(1), e12216. https://doi.org/10.1111/cpsp.12216.
• Overall, N. C., Fletcher, G. J. O., Simpson, J. A. & Sibley, C. G. (2009). Regulating partners in intimate relationships: The costs and benefits of different communication strategies. Journal of Personality and Social Psychology, 96(3), 620–639. https://doi.org/10.1037/a0012961. Direct strategies felt unsuccessful but produced change over a year; soft indirect ones did not.
• Impett, E. A., Gable, S. L. & Peplau, L. A. (2005). Giving up and giving in: The costs and benefits of daily sacrifice in intimate relationships. Journal of Personality and Social Psychology, 89(3), 327–344. https://doi.org/10.1037/0022-3514.89.3.327. Giving in to avoid conflict was linked to lower well-being and relationship quality over time.
• Schrodt, P., Witt, P. L. & Shimkowski, J. R. (2014). A meta-analytical review of the demand/withdraw pattern of interaction and its associations with individual, relational, and communicative outcomes. Communication Monographs, 81(1), 28–58. https://doi.org/10.1080/03637751.2013.813632. Seventy-four studies, 14,255 people; the pattern is moderately linked to worse outcomes whichever partner demands.
• Hahlweg, K. & Richter, D. (2010). Prevention of marital instability and distress: Results of an 11-year longitudinal follow-up study. Behaviour Research and Therapy, 48(5). https://doi.org/10.1016/j.brat.2009.12.010. Couples who took a five-session communication programme separated or divorced at 27.5% against 52.6%.
• Rathgeber, M., Bürkner, P.-C., Schiller, E.-M. & Holling, H. (2019). The efficacy of emotionally focused couples therapy and behavioral couples therapy: A meta-analysis. Journal of Marital and Family Therapy, 45(3). https://doi.org/10.1111/jmft.12336. Thirty-three trials, 2,730 participants; g = 0.60 at post-test.
• VitalSmarts (2016). Costly Conversations: Why the Way Employees Communicate Will Make or Break Your Bottom Line. Survey of 1,025 people; and Crucial Learning (2022), Costly Conversations, survey of 1,100 people. Vendor surveys.
• Bravely (2019). Understanding the Conversation Gap. Survey of 500+ US employees; seven in ten avoiding a difficult conversation at work. Vendor survey.
• Also read: Conflict practice: the guide · How to practice a difficult conversation · The same fight over and over: four conflict patterns

Last updated: November 2026.