Should You Run This Group Online Or In Person?

You have a group to run in January, a small budget, and two options: book a room or buy a Zoom licence. The room costs money and limits who can get there. The licence costs almost nothing and reaches people who cannot travel. Most teams treat this as a values question and argue about it for weeks.

It is not a values question. It is a question about who you are trying to reach and whether your group has a task. Get those two right and mode becomes a logistics decision. Get them wrong and neither mode will save you.

The variable that actually predicts whether the group works

The 2022 qualitative meta-synthesis in BMC Health Services Research on how people experience social prescribing found that participants described benefit extending well beyond social contact — restored purpose, meaningful participation, being useful. Structured, purposeful group activity came out looking more effective than contact alone.

That finding is about content, not venue. A choir, a repair café, a peer-led cooking session and a community garden all have a task. A “friendship group” and a “virtual coffee morning” do not. The evidence that discriminates between good and bad groups is mostly about task, and it applies identically online and off.

So before you choose a mode, write down what people will be doing in the third hour of the third session. If the honest answer is “talking to each other”, you have a design problem that no venue fixes.

What this means in practice Choose the activity first, then let the activity choose the mode. Some tasks are genuinely co-located — cooking, gardening, repair, anything with tools or food. Some are genuinely portable — book groups, language practice, craft-along sessions, peer support with a curriculum. Deciding mode before activity is how you end up with a task that fits the room instead of the people.

What the evidence supports, and how firmly

Claim	Evidence status
Structured, purposeful group activity outperforms unstructured contact	Moderate. Qualitative meta-synthesis, 2022; consistent with participant accounts but not a controlled comparison
Social prescribing referrals to group activity reduce loneliness	Weak to moderate. The 2021 review in Perspectives in Public Health found all nine included studies reported positive individual impacts, and three reported reduced GP, emergency, social work or inpatient use — but these are small and largely uncontrolled
Social prescribing improves self-esteem and confidence	Moderate for direction, weak for size. The 2021 systematic review in IJERPH reports gains but notes limited trial evidence and heterogeneity across programmes
Prosocial roles help lonely older adults	Being tested properly. The HEAL-HOA dual randomised trial published in The Lancet Healthy Longevity in November 2024 is one of very few RCTs in this field rather than an uncontrolled evaluation
Digital delivery is equivalent to in-person delivery for loneliness outcomes	Not established. There is no body of comparative trial evidence to draw on. Anyone telling you otherwise is extrapolating

That last row is the important one. If a funder asks you to justify going online because “the evidence says it works just as well”, the honest answer is that the comparison has not been done at the quality you would need. You are choosing on reach and cost, not on effect size.

Where in-person has a real claim

Physical shared space does something that a video call does not, and the argument for it is about infrastructure rather than format. Eric Klinenberg’s work on social infrastructure argues that libraries, parks and other shared physical places measurably shape how much social contact happens in a neighbourhood, and Ray Oldenburg’s account of third places made the case forty years ago that informal public hangouts are where unplanned association actually occurs. A room in a familiar building generates encounters your programme did not schedule. A Zoom room generates none.

Proximity also matters at the smallest scale. The AARP Foundation’s 2018 survey of 3,020 adults aged 45 and older found 33% of those who had spoken to their neighbours were lonely, against 61% of those who never had. In-person groups produce a walkable network. Digital groups produce a dispersed one, and dispersed networks do not help at 8pm on a bad Tuesday.

Where digital has a real claim

Reach, and only reach — but reach is often the binding constraint. Digital delivery is the only route to people who cannot travel, people on shifts, carers who cannot leave the house, and anyone whose affinity is too rare to fill a room in one town.

Be careful about the remote-work case, though. Gallup’s 2024 global workplace data found 25% of fully remote employees reported loneliness a lot of the previous day, against 21% hybrid and 16% fully on-site. That looks like an argument for pulling people into rooms. But the 2024 cross-sectional study of healthcare workers distinguishes workplace isolation from loneliness as separate constructs with different correlates, and found perceived social support moderating the relationship between remote working and well-being. The December 2024 Scientific Reports analysis makes a parallel point about isolation and loneliness varying in their relationship by age. A digital session aimed at the wrong construct will show no effect and you will blame the mode.

The gallery problem

In any group above about eight people, a proportion attend without ever being individually addressed. Online this is nearly invisible: they join, camera off, name in the participant list, and your register counts them as engaged. In a room, at least someone notices them standing by the coats.

Call it the gallery problem. It matters because Harvard’s Making Caring Common found in 2021 that about half of lonely young adults said no one had taken more than a few minutes recently to ask how they were doing in a way that felt genuine. Your group can reproduce exactly that experience while reporting full attendance.

The fix is procedural, not technological. Cap breakout groups at four to six. Assign one person per session whose only job is to speak to every attendee by name. Track “was spoken to individually” separately from “attended”. If you cannot do this online with your current staffing, that is a real argument against online — not because screens are cold, but because you do not have enough facilitators.

A five-step choice

1. Define the task. What are people making, learning, fixing or organising?
2. Name the population and their travel ceiling. If most cannot reliably get 20 minutes from home, digital is not a preference, it is the only option.
3. Decide which construct you are targeting — isolation, loneliness, or workplace isolation — and pick your instrument before you pick your venue. The 2023 review in BMC Public Health identifies inconsistent measurement as the main barrier to comparing findings, and it will wreck your own year-on-year comparison too.
4. Check facilitator ratio against the gallery problem. One facilitator per six attendees, either mode.
5. Only then choose the mode. If you are tempted by hybrid, staff both rooms separately or do not do it. A hybrid group with one facilitator is an in-person group with spectators.

What this does not solve

Nothing above changes who shows up. Both modes recruit people who already went looking — through a link worker, a poster, a search. The National Academies concluded in 2020 that roughly a quarter of adults aged 65 and over are socially isolated, and the most isolated of them are not on your mailing list in either format. Reporting on American Time Use Survey data in early 2024 showed face-to-face socialising declining before the pandemic and continuing to decline after it, which means the ambient supply of casual contact your group sits inside is shrinking regardless of how you run it.

Mode choice is a question about the 40 people who respond. Reach is a different piece of work, and it is almost always the harder one.