World Café Method: Hosting Rounds, Harvesting, and Analyzing the Output

The World Café is a structured method for hosting dialogue with a large group by breaking it into small café-style tables, rotating participants across several timed rounds, and deliberately cross-pollinating ideas as people carry threads from one table conversation into the next. It was developed in 1995 by Juanita Brown and David Isaacs at a meeting in Mill Valley, California, where rain forced a planned large-circle dialogue indoors; the group improvised by talking in small clusters around tables covered with paper “tablecloths,” rotating between them, and found the resulting cross-table conversation surprisingly rich. Brown and Isaacs then spent several years running action research across multiple countries to formalize what had worked into seven design principles and a repeatable format, later described in their book The World Café.

Most write-ups of the World Café stop at facilitation mechanics — how to set up tables, time rounds, and keep energy up. That covers the hosting half of the method but skips the other half: World Café is also used as a genuine qualitative data-collection method in research, participatory evaluation, and needs assessment, and treated that way it needs the same discipline as any other qualitative method — a deliberate harvesting plan, an explicit analysis approach, and honesty about what the resulting data can and can’t support. This guide covers all three: the round structure, the harvesting step that captures table talk before it evaporates, and how the harvest actually gets turned into analyzable findings.

The Seven Design Principles

The World Café Community Foundation, which stewards the method, defines seven integrated design principles that distinguish a World Café from an unstructured group discussion: clarify the context, create hospitable space, explore questions that matter, encourage everyone’s contribution, connect diverse perspectives, listen together for patterns and insights, and share collective discoveries. The first six shape how the rounds are run; the seventh — sharing collective discoveries — is what harvesting exists to do, and it’s the principle a purely facilitation-focused write-up tends to shortchange.

How the Rounds Work: Tables, Hosts, and Rotation

A World Café session seats participants in small groups, typically four to five people per table, for a series of rounds — three is the most common count, each running roughly 20-30 minutes, though both figures are design choices tuned to the group size and question set rather than fixed rules. One person per table stays put for the whole session as the table host; everyone else rotates to a new table after each round.

The rotation is what makes the method more than a series of parallel focus groups. When a new set of participants arrives at a table, the host briefly recaps the key threads from the previous round, and the new group builds on that thread rather than starting fresh — carrying fragments of earlier conversations into new combinations of people. Across three rounds, most participants will have contributed to three different table conversations, each seeded by what a different set of people discussed before them.

Question design across rounds is itself a methodological choice. Some designs use the same question in every round to look for convergence — do independently-seeded table conversations arrive at similar themes? Others progress the question from broad and exploratory in round one to specific and action-oriented by round three, deliberately narrowing focus as the group’s collective thinking develops. Either way, the questions are prepared in advance as part of the design, not improvised at the table.

The Harvesting Step: Capturing the Conversation Before It’s Lost

Table conversation is dense and short-lived. Once a round ends and the table’s membership scatters, the specific phrasing, half-finished ideas, and disagreements that were said out loud are gone unless something captured them in the room — reconstructing them afterward from memory or a facilitator’s general impression is not a substitute, and this is the step research use of the method most often shortchanges. Harvesting is the deliberate design of how table talk becomes recorded, shareable output, and it typically layers several capture methods rather than relying on one:

• Paper tablecloths. Tables are often covered with large sheets of paper and set with markers or crayons, and participants doodle, jot keywords, or sketch connections during the conversation itself. This is simultaneously a facilitation device — it keeps hands busy and lowers the inhibition some participants feel about speaking in a group — and the first, most immediate layer of data capture.
• Table host summaries. After the final round, each table host gives a brief verbal report to the whole room, or completes a short written summary sheet, distilling what that specific table’s sequence of conversations produced.
• A facilitated harvest conversation. The whole group reconvenes and discusses patterns across tables together — not each table reporting in isolation, but a collective conversation about what recurred, what diverged, and what surprised people. This is often where a graphic recorder produces a live visual synthesis of the emerging patterns.
• Audio or written records agreed in advance. For research use specifically, relying on tablecloths and a closing conversation alone is usually too thin. A study-grade World Café plans its capture before the session starts — assigning a scribe per table, recording table conversations with participants’ informed consent, or requiring legible written table notes — rather than deciding after the fact that better records would have been useful.

From Harvest to Data: Analyzing World Café Output as a Research Method

Once harvested, the tablecloths, table-host notes, and harvest-conversation record become the dataset, and researchers who use World Café deliberately (rather than purely as a facilitation exercise) analyze it the way they would any other qualitative data. The most common approach is thematic analysis: coding the harvest artifacts to identify recurring themes, then examining how those themes built or shifted as the same question moved through different table combinations across rounds. A lighter-weight content analysis — counting how often specific ideas or keywords recur across tables and rounds — is also common where the goal is a quick synthesis rather than a deep interpretive account.

One distinction matters more for World Café than for most qualitative methods: the data is inherently collective, not individually attributable. In a focus group, a transcript can usually still be coded by speaker. In a World Café, once several different people have cycled through a table across rounds and the record is a shared tablecloth or a host’s summary rather than a speaker-attributed transcript, there is generally no way to trace a given idea back to the individual who first said it. That rules out any analysis that needs individual-level attribution — you can’t report “participant 4 said X” — and it’s a limitation to state plainly rather than paper over. The practical consequence is that analysis usually happens at the level of the table (as the unit that produced a given piece of harvest) and then across tables, rather than at the level of the individual participant.

The harvest conversation itself doubles as a form of real-time member checking: when a facilitator or researcher reflects the emerging cross-table patterns back to the same room that generated them, participants can confirm, correct, or complicate that synthesis before the session ends — a validation step that a method relying only on post-hoc researcher coding doesn’t get for free.

For a large multi-table World Café — dozens of tables, or several sessions run across different sites on the same question — the volume of tablecloths, notes, and transcripts can exceed what’s practical to code by hand. At that scale, teams typically bring in qualitative analysis software to organize and code the harvest material, the same tooling used for interview or focus-group transcripts; see CAQDAS workflows for how that coding process works in practice.

Where World Café Sits Among Group-Based Qualitative Methods

World Café is one of several structured ways to generate qualitative data from a group rather than an individual, and it’s easy to conflate with the others. The differences are mostly about what each method optimizes for:

• Vs. a focus group: a focus group is one conversation with a fixed group and a moderator steering toward the research questions, and comments can usually be attributed to individual speakers. A World Café runs many small parallel conversations with rotating membership, and the output is collective rather than individually attributable.
• Vs. the nominal group technique: NGT deliberately suppresses open discussion at key steps — silent idea generation, then individual voting — specifically to prevent groupthink and produce a ranked, scored list. World Café is intentionally the opposite: it’s built for social, cross-pollinating conversation and shared sense-making, not a ranked output.
• Vs. the Delphi method: Delphi’s rounds are anonymous, asynchronous, and run over days or weeks, with a facilitator statistically aggregating responses between rounds toward measured consensus. World Café’s rounds happen face-to-face in a single sitting, participants are not anonymous to each other, and the goal is emergent shared patterns rather than a converged consensus score.

Limitations to Report Honestly

Used as a research method, World Café has real constraints that belong in any methods section that reports it, not just in a facilitator’s post-event notes:

• No individual attribution — covered above, and it rules out participant-level analysis by design, not by omission.
• No widely-adopted, standardized reporting checklist. Methods with a large formal research literature, like the Delphi method, tend to have consensus reporting conventions; World Café, coming out of organizational-development and facilitation practice rather than a single research tradition, doesn’t have an equivalent widely-adopted standard, so studies vary considerably in how much methodological detail they report.
• Format sensitivity. Question wording, room setup, who hosts each table, and group energy on the day all shape what a Café produces. That’s a feature for engagement but a limitation for replicability — the same question set run by a different facilitator with a different room may not surface the same patterns.
• Scale of participation is not statistical generalizability. A Café with a large number of people in the room can feel like a large sample, but it’s still a method for depth and breadth of collective sense-making within that specific group, not a way to produce findings that generalize to a wider population the way a probability sample would; see sampling methods for that distinction.

A Practical Design Checklist

• Table size: 4-5 participants per table, one designated as table host who stays for the whole session.
• Round count and length: three rounds is typical, roughly 20-30 minutes each — adjust to group size and question depth, not a fixed rule.
• Decide the harvesting plan before the session, not after: tablecloths plus markers at minimum; add a per-table scribe or consented audio recording if the output needs to support formal analysis.
• Write the question set in advance, and decide deliberately whether it repeats each round (testing for convergence) or progresses from broad to specific.
• Budget real time for the harvest conversation at the end — it is not a five-minute wrap-up, it’s where principle seven, sharing collective discoveries, actually happens.
• Decide the analysis approach (thematic coding, content analysis, or a lighter synthesis) before the session, so the harvest is captured in a form that analysis can actually use.

Frequently Asked Questions

How many rounds does a World Café have?

Three rounds is the most common design, but the count is a design choice rather than a fixed rule — it depends on group size, how many table combinations are useful, and how much time is available. What matters more than the exact number is that participants rotate between each round and a table host stays behind to carry the thread forward.

What’s the difference between a World Café and a focus group?

A focus group is a single, fixed-membership conversation led by a moderator, with output that can usually be attributed to individual speakers. A World Café runs several small parallel conversations with rotating membership across multiple rounds, and its output — tablecloths, host summaries, a harvest conversation — is collective rather than individually attributable. See Focus Groups in Research for the focus-group side of that comparison.

How is World Café output analyzed as research data?

The harvest artifacts — paper tablecloths, table-host notes, and the recorded harvest conversation — are treated as the qualitative dataset and coded the way any other qualitative data would be, most often through thematic analysis, usually at the level of the table rather than the individual participant, since individual comments generally can’t be attributed once table membership has rotated.

Can the World Café method be used as a standalone research method, rather than just a facilitation technique?

Yes, but it requires treating it that way from the design stage — a deliberate harvesting plan agreed before the session, an explicit analysis approach, and honest reporting of its limitations (no individual attribution, no standardized reporting checklist, sensitivity to facilitation and room conditions). Used purely as a facilitation exercise with no harvesting plan, the output is useful for the room but isn’t research-grade data.

What is “harvesting” in the World Café method?

Harvesting is the deliberate capture of table conversation before it’s lost when a round ends and table membership scatters — layering paper tablecloths, table-host summaries, and a facilitated whole-group harvest conversation (sometimes with live graphic recording) so that what was said becomes recorded, shareable output rather than a fading impression.

Related CASRAI Resources

World Café sits alongside other structured group methods for generating qualitative data. See Focus Groups in Research and the nominal group technique for two other group formats with different tradeoffs, and the Delphi method for the asynchronous, anonymous-consensus counterpart. For analyzing whatever a Café’s harvest produces, Thematic Analysis: A Step-by-Step Guide covers the coding process in depth, Member Checking in Qualitative Research covers the validation step the harvest conversation partly substitutes for, and CAQDAS Workflows covers the software options once harvest volume outgrows manual coding. Qualitative Research provides the broader methodological context, and Sampling Methods covers how table/session participants are typically recruited.