Virtual vs. In-Person Roleplay Training: What Changes and What Doesn't

Our second pilot of Operation Aetherfall ran seven and a half hours at one table. Six players, a pile of dice, a 90-minute debrief, and the kind of shoulder-to-shoulder energy you only notice once it's gone. If a client asked us to run that exact day over Zoom, the honest answer would be no. The slightly longer answer is that we could still get them the same learning outcomes – in a different shape.

That's the short version of what the research says about virtual versus in-person soft skills training. The parts that carry the learning survive the move to video: structured practice on authentic dilemmas, and a well-run debrief afterward. Controlled studies keep finding virtual delivery matches or beats in-person delivery on skill outcomes. What degrades is everything around the practice – the nonverbal channel, the informal conversations that build trust, the group's energy, and the facilitator's ability to read who's struggling. Going virtual doesn't lower the ceiling. It hands you a bill: smaller groups, shorter blocks across more days, 30-40% fewer learning objectives, and safety practices an in-person room gives you for free.

I've worked remotely for over a decade, and the training we build runs on a tabletop RPG – a format born at physical tables. I wanted the evidence to say video costs nothing. It doesn't. It says something more useful: exactly what video costs, and exactly what buys it back.

What survives the move to video?

The core of experiential training – practice plus debrief – transfers better than most people expect. A randomized trial of 120 healthcare professionals found that practicing with a virtual standardized patient produced greater improvement on three of four motivational interviewing skill scores than traditional academic study, with the gains holding over time. Closer to conflict work specifically, a study of an LLM-based conflict simulation called Rehearsal found participants cut their use of escalating competitive strategies by an average of 67% and doubled their cooperative strategies compared to lecture-only training – and the improvement showed up afterward in an unaided, live conflict. Practice over a screen still counts as practice.

The debrief holds up too. A randomized trial in immersive VR compared facilitator-guided debriefing against self-guided: both improved teamwork skills, and while participants rated the facilitated version higher on quality (a mean difference of 1.4 points), workload and motivation were the same across groups. The frameworks that structure a good in-person debrief – reactions, analysis, summary – run over video without much modification. What actually threatens the virtual debrief isn't quality. It's omission: roughly half of virtual simulations skip the post-simulation debrief entirely. The cheapest way to beat the average is to actually run one.

And the fictional frame doesn't care where you sit. Simulation research consistently finds that psychological fidelity – whether the scenario evokes real pressure and real decisions – predicts learning, while physical fidelity barely registers. In one study, military cadets practicing negotiation performed the same in a military-skinned virtual scenario as in a civilian one on 27 of 28 measures. A scenario that works because of its pressure rather than its set dressing survives the loss of the room.

What degrades over video?

Four things, and they're all in the human channel.

The nonverbal layer thins out. Video carries words well and everything else badly – researchers describe computer-mediated tools as having "limited capacity in transferring verbal and nonverbal cues between users." Posture, micro-reactions, who just leaned back from the table: gone or garbled. For training built on reading people, that's a real loss, and it hits the facilitator hardest.

Informal connection disappears. There's no hallway before a Zoom call. Trust research on virtual teams finds that cognitive trust – confidence in a colleague's competence – develops fine over video, while affective trust, the emotional kind that lets people be vulnerable in front of each other, forms far less readily. Conflict practice runs on the second kind. Patrick Lencioni, whose whole model rests on vulnerability-based trust, titled a reflection on his own distributed team "Virtual Teams are Worse than I Thought."

Energy drains faster, and the evidence here is physiological, not vibes. A 2023 study wired 35 participants with EEG and ECG and had them attend the same 50-minute lecture by videoconference and in the lecture hall; the videoconference produced measurably greater brain-level fatigue. Eye-tracking studies of virtual consultations show cognitive fatigue emerging in the first 10-12 minutes, against 45-60 minutes in person. Stanford's Jeremy Bailenson traced the mechanisms: close-range eye gaze from a grid of faces, a mirror of your own face all day, a body pinned in frame, and the extra work of producing and decoding nonverbal signals on purpose. Worse for our purposes, a 2024 study found that fatigued participants conform more – dissent costs energy they no longer have. In conflict training, that's the exact failure mode: a room of tired people politely agreeing.

One honest complication: a 2024 replication study (125 people, 945 meetings) found that post-pandemic, video meetings are no longer inherently exhausting – meetings under 44 minutes were less tiring than other meeting types, and only the boring ones wore people out. Video doesn't doom a session. Length and passivity do.

Safety signals get harder to read. Power dynamics become more pronounced on video, silence feels more awkward, and a participant who's quietly overwhelmed can mask it, reframe the camera, or vanish behind "technical difficulties." In person, a facilitator catches the shift in posture two seats away. On video, there's often nothing to catch. The same screen distance also dampens emotional intensity – participants can intellectualize a conflict scenario instead of feeling it, which produces tidy discussion and no behavior change.

How do you buy back what video takes?

With structure. Every degradation above has a documented compensation, and none of them requires special technology – they require accepting that a virtual training day is a different animal from an in-person one.

What degrades	What compensates	The numbers to hit
Attention and energy	Shorter blocks, more breaks	60-90 minute sessions; a 5-10 minute break every 20-30 minutes; never past 90 without one
Content absorption	Fewer objectives, more days	Cut learning objectives 30-40%; run multi-day, not a single-day marathon
The facilitator's read on the room	Smaller groups, explicit check-ins	3-5 people per breakout; no more than 10 in a facilitated debrief
Equal participation	Structured turn-taking	Round-robin with a pass option; planned interaction every 3-5 minutes
Safety signals	Longer, explicit pre-brief and camera norms	15-20 minutes of pre-brief; self-view turned off; camera-off allowed as an escape valve
Trust and informal connection	Deliberate relationship time	30-40% of early sessions on relationship building before conflict content

The biggest single change is the shape of the day. One comparative study found a virtual program hit the same skills-based criterion as an 8.5-hour in-person workshop using five synchronous hours segmented across multiple days – 40% less live time, same outcome. Meanwhile, a facilitator who watched a full-day online cohort limp through its last block asked, in print, "is that final 90 minute session even worth it!?" Our seven-and-a-half-hour pilot day would become something like three two-hour sessions across a week, with reflection work in between. Not worse. Different. More on structuring the day in game-based training day logistics.

The safety work also moves from implicit to explicit. In person, rapport builds in the coffee line. On video you replace it with a longer pre-brief, a stated fiction contract, normalized mistakes, and agreed signals for pausing a scene – the toolkit we cover in safety tools for workplace roleplay.

When is virtual actually the better format?

More often than nostalgia for conference rooms admits.

For distributed teams, the honest comparison isn't Zoom versus a shared room. It's Zoom versus nothing, or Zoom versus flying eight people to one city. Virtual delivery removes travel cost entirely and lets the whole team train together, including the people a physical offsite always seems to exclude.

For quieter participants, video can be a genuine upgrade. One research summary puts it flatly: "when done well, virtual environments can actually increase psychological safety for introverts, people with disabilities, and those balancing caregiving responsibilities." Chat channels, anonymous polls, and structured turns give people ways to contribute that don't require winning a fight for airtime. A leader of an international team noticed that when everyone went remote, the members who had always worked from home "started to speak up and contribute more. Having everyone on video was a great equalizer." There's parallel evidence that asynchronous online discussion improves gender equality of participation relative to face-to-face.

And for high-stakes practice, distance can enable what proximity prevents. For teams with power imbalances or a real conflict history, researchers note that virtual distance provides a physical safety that makes psychological risk-taking possible – people will practice a confrontation on a screen they'd never attempt across a table. The double edge is that the same distance that makes practice feel safe can make it too safe. The fix isn't the room; it's scenario pressure plus repetition. Virtual platforms make retrying a hard conversation cheap, and in one medical training study, rapid-cycle practice with immediate feedback produced 98% skill proficiency that still held six months later.

So which format should you choose?

Start with trust. If the team is new to each other or trust is thin, start in person – the distributed-team literature is blunt that new teams should meet physically first, and virtual trust-building timelines can't be compressed without sacrificing the training. If in-person isn't possible, budget 30-40% of your early virtual sessions for relationship building before any conflict content, and treat that as curriculum, not warm-up.

Then ask what you're training. Process skills – interest-based questions, de-escalation language, structured listening – train well over video, arguably better, because reps are cheap and feedback is immediate. Unscripted, emotionally hot conflict work wants a shared room, because it runs on affective trust and full-channel communication.

If the stakes are high and the team is spread out, the research points to hybrid: an in-person kickoff of two to five days to build the trust foundation, weekly virtual practice sessions for six to twelve weeks, and quarterly in-person time for the hardest scenarios. That sequence uses each format for the thing it's actually good at.

We'll keep running Operation Aetherfall at physical tables whenever we can get one, because the energy is real and I like the sound of dice. But the thing that changed our players' behavior was never the table. It was the pressure and the debrief – and both of those travel.