The room is not a measurement

Attendance is reported because it is easy to count, not because it describes anything useful.

Questions need design

Open-floor questions reliably produce long statements from the same few people. Collecting questions in writing, or having the chair select, changes the quality of a session more than almost any other intervention.

Seating decides participation

Rows facing forward tell the room it is there to receive. Any arrangement where people can see each other produces more discussion, and the effect is immediate and large.

Give them something to do

Even two minutes of turning to a neighbour and comparing reactions changes what people retain. It feels like a gimmick to organisers and is the single most reliable retention device available.

Name badges carry information

A badge with a name and nothing else creates awkwardness. A badge with one line about what the person works on starts conversations that would not otherwise happen.

Ask afterwards, specifically

A feedback form asking for a rating produces noise. One asking what the person will do differently produces answers you can act on, and far fewer of them, which is an improvement.

Accessibility is programming

Captioning, step-free access, a quiet room and a published schedule with real timings determine who can attend at all. These are usually handled as compliance items late in planning, when they are expensive, instead of as design decisions when they are cheap.