This mixed methods, multiple case study examined how communication medium influenced task performance and early interpersonal development in newly formed small groups. Data was collected from 24 volunteer participants arranged into six groups of four. Participants were experienced professionals recruited from professional or organisational settings. Participation was voluntary and subject to informed consent. No demographic screening or matching criteria were applied. Participants were randomly allocated to ad hoc groups to reduce systematic bias arising from pre-existing familiarity.

Two groups completed the study in each of three communication conditions: face-to-face interaction, Microsoft Teams video conferencing, and synchronous text chat using Microsoft Teams. Face-to-face sessions took place in designated meeting rooms; remote participants joined the relevant Microsoft Teams environment. Technical arrangements were tested in advance, and every group received the same written and verbal instructions and followed the same session structure.

Each session comprised two minutes for arrival and group orientation, four minutes of facilitator instructions, an eight minute collaborative task, and up to 15 minutes for an individual post-task questionnaire. Participants remained in the study environment for the full eight minute task period, even if they completed the task early. The facilitator provided no further prompts after the task began and did not direct the use of any remaining time. This allowed spontaneous task-related and non-task-related interaction to occur.

The task was a standardised collaborative problem solving exercise. Six mountain landscape images were each divided into four interlocking puzzle pieces. The images were visually similar, with mountain and sky features in the upper pieces and landscape foregrounds in the lower pieces. For each group, puzzle pieces were selected and distributed so that one complete four piece image could be reconstructed from the available set, alongside visually plausible distractor pieces. Each participant received three pieces and held one piece required for the correct solution. The task could therefore not be completed independently: participants had to share information, compare their pieces, negotiate possible combinations and collectively identify the single correct four piece image. Task versions and piece selections were varied between groups to reduce the risk of prior knowledge. The primary objective measures were whether the correct solution was identified and the elapsed time to solution, recorded using a stopwatch.

Group interaction was captured according to communication condition. Face-to-face sessions were audio recorded and transcribed using Google Voice Recording, video-conference sessions were recorded in Microsoft Teams, and text-chat sessions were retained as Microsoft Teams chat logs. Recordings and logs were transcribed and anonymised. The transcripts were coded to quantify task-related and non-task-related communicative exchanges and to examine communication patterns, collaboration, conflict, cohesion, interpersonal disclosure and spontaneous social interaction.

Immediately after the task, participants individually completed a structured questionnaire containing scaled, ranking and open-text questions. Measures included perceived effectiveness of the assigned communication medium on a 1–10 scale; estimated task focus during the eight-minute period; relationship familiarity before and after the task on 1–10 scales; trust in other group members before and after the task on 1–10 scales; whether participants had learned anything new about or from another participant, with opportunities to explain their responses; and a ranking of five communication media: instant messaging, video conference, face-to-face interaction, voice conference and virtual reality, by expected effectiveness for the task.

The resulting dataset comprised task completion outcomes and times, anonymised audio/video transcripts and chat logs, coded interaction counts, scaled questionnaire responses, medium rankings and open text participant reflections. Quantitative data was compared across the three communication conditions using descriptive statistics, one-way analysis of variance and Tukey post-hoc analysis where appropriate. Qualitative responses and interaction data were analysed thematically and integrated with the quantitative results through cross-case comparison. All participant records were anonymised and handled in accordance with the study’s institutional ethical approval and data-protection requirements.