How to Document a Citizens' Assembly or Public Dialogue With Many Tables

A practical guide to capturing what dozens of citizen tables say and turning it into one report that holds up to public scrutiny, from the classic "theme team" to keeping a synthesis traceable and honest about disagreement.

Capturing what was said across dozens of tables at a citizens' assembly, and turning it into one report the public will accept, is really two jobs. The first is getting the raw material off every table without losing the reasoning behind it. The second is combining all of it into themes that people can trace and challenge. The second job is where these events succeed or fail. The report is the product, and a report that looks cherry-picked discredits the whole exercise. The old name for the people who do that second job is a theme team.

The theme team, the oldest answer to this problem

Large public dialogues have wrestled with the many-tables problem for decades. One well-documented approach came from AmericaSpeaks, a nonpartisan US nonprofit that ran a format it called the 21st Century Town Meeting from the late 1990s. Citizens sat at round tables of ten to twelve, each with a trained facilitator keeping the conversation balanced. A scribe at every table typed the group's points into a networked laptop, and those notes streamed to a separate group working out of sight: the theme team. Their job was to read across all the tables at once and pull out the strongest themes while the event was still running, which were then shown back on large screens for the whole room to react to and vote on with handheld keypads. The method is documented on Participedia. The same format was used in New Orleans after Hurricane Katrina to help shape the Unified New Orleans Plan, with thousands of residents, many of them displaced to other cities, joining by satellite link.

The point of the theme team wasn't speed. It was legitimacy. Because the synthesis happened in the room and got shown back for the room to correct, no one could later claim the organisers had quietly rewritten what people said. That is the standard any documentation of a citizens' assembly is really being held to, whether or not there's a keypad in sight.

Why the synthesis is the part that gets attacked

For an ordinary workshop, a rough write-up is fine. For a citizens' assembly or a public consultation, the synthesis is a political document. Elected officials, journalists, and the participants themselves will ask whether it really reflects what the room said, or whether the organisers heard what they wanted to hear. A synthesis that smooths a divided room into a tidy consensus is easy to attack, and once its fairness is in doubt the whole process loses its point. So the work isn't only to write a good summary. It's to write one that survives someone trying to poke holes in it.

Keep every theme traceable back to a table

The single most useful habit is provenance. For every theme or claim in the final report, you should be able to say which tables it came from and roughly how many. That turns "there were concerns about traffic" into something a skeptic can check against the raw notes instead of taking on trust.

• Number the tables and label every note, photo, and recording with its table before the event starts. An unlabelled pile of notes can't be traced back to anything, and relabelling after the fact is where doubt creeps in.
• Tag each theme in the synthesis with the tables that raised it, and how many did. A concern forty tables voiced and a concern two tables voiced are both real, but they are not the same finding, and collapsing them hides the difference.
• Keep the raw table notes as an appendix, not just the polished summary. Being able to show your working is what makes the summary credible when someone questions a specific line.
• When you compress a long discussion into a single line, keep a pointer back to the fuller note. The reasoning is usually the first thing lost in compression and the first thing a critic asks to see.

Report disagreement instead of averaging it away

The most useful information in a public dialogue is often where citizens disagreed, not where they agreed. A synthesis that reports only the majority view hides exactly what decision-makers need to see. Say where the room split, roughly how the split fell, and the reasoning on each side. Minority positions reported honestly make a report more credible, not less, because they show the synthesis wasn't steered toward a predetermined answer.

This is part of why running these processes has grown into a studied field with its own standards. In its 2020 report Catching the Deliberative Wave, the OECD counted 289 representative deliberative processes across its member countries between 1986 and October 2019, and good-practice guidance for them consistently stresses transparency about how conclusions were reached, not only the conclusions themselves. A report that shows its reasoning and its disagreements is doing what that guidance asks.

Read the themes back to the room before people leave

The 21st Century Town Meeting fed themes back on screen for comment and a vote in the same session, and that step is worth copying even without the technology. Read the emerging themes back to participants before they go home. Ask whether anything was misheard or left out. A correction made in the room, while people are still there, is worth far more than one made a week later over email, and it means no one can honestly say the synthesis was written behind closed doors.

Capturing the raw material at scale

Everything above is about the synthesis. The capture step underneath it is the labour-intensive one, and there is no free version of it. A scribe at every table takes bodies out of the conversation. A professional note-taking service adds a cost that scales with the number of tables, since you need roughly one note-taker per table plus a team to pull it together. Recording each table on its own device captures everything but leaves someone hours of audio to transcribe and read back afterwards.

One example of software built for the parallel-capture problem is RoomRadar. Each table uses a phone as its microphone, with participants joining by scanning a QR code in a browser, no app to install and no account to create. Every table is transcribed live at the same time, grouped into themes, and turned into a written report automatically, and the raw audio isn't stored, only the text and the structured analysis. More on the table setup is here.

Whichever route you take, the capture method mainly decides how much raw material you have to work with. It doesn't decide whether the synthesis built on top of it is defensible. That still comes down to the traceability and the honesty about disagreement described above, which are the same whether the notes were typed by a scribe or transcribed by a machine.

Decide these before the day, whatever method you use

• Who owns the synthesis, and are they seen as neutral? If the same body being consulted also writes the summary of what citizens said, expect the result to be questioned. A visibly independent synthesiser buys credibility that no amount of careful wording can.
• How will minority views be reported? Decide this before you see the results, so the answer isn't quietly shaped by which way the room happened to go.
• What happens to the raw notes and any recordings, and for how long? Decide retention up front, write it down, and be ready to tell participants.
• What exactly are you telling participants is being captured, and why? Say it plainly at the start, not in fine print afterwards.

FAQ