A Workshop That Produces Something

Most events called workshops are small conferences. People present finished work, an audience listens, questions are asked, everyone leaves.

That is a perfectly good format and it is not a workshop. A workshop is for doing something together that could not be done separately, and designing one requires deciding what that something is. For a concrete operational example of how a workplace tool approaches a related coordination problem, see read more.

Decide the output first

What exists at the end that did not exist at the start?

A shared research agenda — an agreed statement of what the open questions are.

A joint paper or a special issue.

Improved drafts, where the point is that everyone's work is better.

A method or a standard the participants agree to use.

A network that will still be functioning in a year.

"Discussion" is not an output. If the honest answer is that people will talk and something may come of it, that is a small conference, and calling it that sets the right expectations.

Format follows from the output

If the output is improved drafts: papers circulated in advance, everyone reads everything, each session is discussion rather than presentation. The author speaks for five minutes, a discussant for ten, and the room for thirty. This is the highest-value format and it requires participants who will actually read.

If the output is a shared agenda: structured sessions producing written material, with someone assigned to capture it. Free-form discussion produces a good afternoon and no document.

If the output is a joint publication: the structure and the authorship arrangements should be settled before the workshop, not discovered during it. See authorship.

If the output is a network: the design problem is contact time and mixing, not content. Fewer sessions, more unstructured time, deliberate seating.

Size and selection

Small. Fifteen to thirty. Above that, discussion stops being possible and it becomes a conference.

Invite for fit, not for status. A workshop works when the participants have something to say to each other about the same problem.

Mix seniority deliberately. A room of established professors produces a different conversation from one that includes people who joined the field last year, and the second is usually better.

Say what you expect of participants in the invitation: read everything, arrive having prepared, stay for the whole event. A workshop where half the room drops in for one session does not work.

Designing the sessions

Longer sessions, fewer of them. Ninety minutes on one paper produces more than three twenty-minute talks.

Reverse the ratio. Five minutes of presentation, forty of discussion — for work everyone has read.

Name a discussant for each item, and tell them well in advance. See being a discussant.

Assign someone to write. Every session that is meant to produce something needs a person whose job is capturing it, and they should not also be chairing.

Build in synthesis time. A session at the end of each day, and at the end of the workshop, where the room agrees what was concluded. This is where the output actually gets made and it is the thing most often cut when the schedule slips.

The mistakes

Papers circulated too late. If they arrive three days before, nobody reads them and the format collapses into presentations. Send them three weeks ahead and remind people.

Too many items. Ten papers in two days is a conference. Five is a workshop.

No output owner. "We'll write something up afterwards" means nobody does. Name the person, before the event.

Every session full. A workshop with no gaps has no conversation, and the conversation is what people came for.

Ignoring the follow-through. The month after is when the output is either produced or abandoned, and it needs a schedule of its own.

The follow-through

Where most workshops fail, having gone well.

Circulate the notes within a week, while people still recognise what they agreed to.

Set the next deadline before people leave. A date for drafts, a date for comments, a date for the compiled version. Agreed in the room, with names attached.

Name who does what. A list of tasks with no owner is a list.

Schedule one follow-up call, at a date agreed in the room.

And be realistic. If the output requires substantial work from busy people afterwards, it will happen only if someone owns it and chases. Frequently the honest design is to make the output something that can be largely finished during the workshop itself.

Practical points

A room where people can see each other. A lecture theatre defeats a workshop before it starts. Tables in a square, or a circle.

Whiteboards, and someone using them.

Food that does not require leaving the building, so the conversation continues.

Somewhere to work individually for the parts that need it.

Accessibility, planned in advance, as with any event. See accessibility.

And a modest budget — workshops are cheap relative to conferences, which is part of their appeal.

The short version

Decide the output first. If it is "discussion", you are running a small conference — which is fine, and call it that.

Format follows from output, and for improved drafts the ratio is five minutes of presentation to forty of discussion.

Fifteen to thirty people, papers circulated three weeks ahead, discussants named in advance.

Someone owns the output, named before the event, with deadlines agreed in the room.

And leave gaps — the conversation is what people came for.

For an independent authoritative reference, consult ACM conference resources for professional guidance on academic events.