Ask almost any deliberation practitioner whether in-person engagement beats online, and you'll get the same instinctive answer: of course it does. Richer conversations, deeper listening, more enjoyment for participants and facilitators alike. It feels obviously true.

There's just one problem. For all the received wisdom that shapes this field, there's remarkably little comparative data to test it. 

We think commissioners of deliberation should be able to make decisions grounded in evidence. So we’ve tested the biggest assumption of all: is in-person really superior to online?

The short answer: we found no evidence that it is.

Between July and November 2025, we ran 36 deliberative workshops on AI in public services. 18 took place in person - in Newcastle, Birmingham and London - and 18 online. Everything else was held constant: the same 3.5-hour design, the same stimulus content and facilitation guide, the same core team of facilitators across both modes, and the same digital platform (Zeitgeist) throughout. In total, 280 participants took part, randomly assigned to workshops and broadly reflective of the UK population.

We then tested whether delivery mode shifted any of four outcomes: participants' rating of deliberation quality, their enjoyment of the session, their sense of how effective the process was, and how far their attitudes to AI moved over the course of the workshop.*

On three of the four outcomes, we found no evidence of a meaningful difference between the two modes of delivery.

Deliberation quality - our composite of whether everyone had the chance to contribute, whether people with different views could explain their perspective, and whether participants listened to one another - scored an average of 4.76 out of 5. This was high enough that there was little room left on the scale for any differences between the online vs in-person modes to pull apart, and we detected no statistically significant difference between them. Participants rated the process highly effective at involving the public in the governance of AI in public services, again with no detectable gap between online and in the room. We also measured how five key attitudes shifted as a result of taking part including: comfort with public sector AI, perceived societal benefit of AI, ability to influence decisions about AI, and trust and knowledge of public sector AI. Overall, attitudes shifted by a similar amount in both formats (on average across all measures we observed a positive shift of +0.39 points on a 1-5 scale, equivalent to ~10%), with no significant difference between them.

The one exception was enjoyment, and here the difference was tangible: in-person workshops scored modestly higher, 9.26 compared to 8.75 out of 10 for online - about half a point (0.51, p=.007), still from a high base in both cases.

For us, the conclusion is clear enough to act on: we found no evidence that a well-designed online workshop on our Zeitgeist platform performs worse than in-person workshops on key measures - and at a lower cost. The data doesn’t prove the two are identical, but our experiment design was sensitive enough to catch the half-point enjoyment gap, so any difference we missed elsewhere is likely smaller still. That modest enjoyment edge will rarely justify the difference in cost.

A few caveats. These results are specific to our design and platform; they may not hold for generic video-conferencing tools such as Teams, Zoom or Google Meet, or for other suppliers. And there are still good reasons to meet face-to-face - sensitive subject matter, say, particular access or support needs, or the need to build trusting relationships with participants. But the headline stands: done well, online deliberation holds its own.

If you're commissioning public deliberation and want to talk about designing it for both quality and value - online, in person, or a blend of the two - we'd love to hear from you. Please email us at [email protected].