Most programme decisions are made on instinct, because almost nobody measures what attendees do with a programme once it exists. PheedLoop, an event platform, did: it analysed 967,701 session check-ins across 411 events.1 The results are useful mainly for how many common worries they put to rest.
Attendees have a fixed appetite
At the median event, an attendee who went to any session attended 5.5 of them, with the middle half of events between 3.3 and 8.0. PheedLoop reports the figure as stable: cutting the data in different ways moved the median only between 5.4 and 7.0.1
And not everyone goes to the programme at all. At the median event, about a quarter of the people who checked in at the door never checked into a single session; at 32% of events, fewer than 60% did.1 PheedLoop notes this figure is approximate, because door check-in is not run equally thoroughly everywhere.
More sessions, less than proportional return
This is the finding with the biggest budget consequence. Sessions attended per person rose with programme size, but slowly:1
Going from roughly 10 sessions to roughly 80 is an eightfold increase in programme for about 2.5 times the attendance per person. Length behaves the same way: single-day events saw 3.7 sessions per person, two- to four-day events 5.4 to 5.9, and five-day events 8.4.1
An illustration using those bands. A two-day conference expects 600 people and is choosing between 30 sessions and 50. If three in four go to the programme, that is 450 session-goers. At 30 sessions and 5.8 each, that is about 2,610 session visits, or 87 per session. At 50 sessions and 7.2 each, it is about 3,240 visits, or 65 per session. The 20 extra sessions add about 630 visits, roughly 32 each. That may still be worth it, but it is a different decision from "more content, more value".
Every extra session costs a room, AV, a speaker's travel and a slot in the printed agenda. Past a point, it mainly competes for attention your attendees have already given to something else.
What barely matters
PheedLoop indexed each session against its own event's typical session, then tested the variables committees argue about most.1
Time of day. 8am sessions indexed at 1.05, 1pm at 1.00, 4pm at 0.91. A modest late-afternoon dip, and no post-lunch collapse.
Session length. From under 30 minutes to over two hours, every band sat between 0.97 and 1.01.
Parallel tracks. Slots running 2 to 12 concurrent sessions indexed between 0.98 and 1.00. Adding tracks did not measurably dilute attendance per session.
Later days. Day 2 ran at 1.03, day 3 at 0.88, day 4 at 0.83: real decay, but smaller than the variation between events.
Nor is attendance as top-heavy as programme debates assume. The top 20% of sessions drew a median 38% of session attendance, and the bottom half 29%.1 Your quieter sessions are carrying more of the programme than they get credit for.
Sizing rooms
Agenda bookmarks are often used to size rooms, and they are right in total and wrong in detail. Median attendance was 90% of bookmarks, but only 28% of sessions landed within 20% of their bookmark count; 40% drew more people than bookmarked them and 18% drew fewer than half.1 Use bookmarks to estimate overall appetite, and keep room assignments flexible enough to swap on the day.
In practice that means holding a larger room in reserve, choosing rooms with partition walls or overflow seating, and on multi-day events using the first day's session check-ins to reassign the second day's rooms. The first day's real numbers beat any number of bookmarks.
Building the programme
Plan backwards from the 5.5. If your attendees will go to about five or six sessions over the event, the question is what those five or six should be for each type of attendee, not how many slots you can fill. Bizzabo's 2026 survey points the same way: asked what most improves attendee personalisation, 40% of organisers named content personalisation, level with personalised on-site activations.2 Relevance is the lever, not volume.
Then look at the quarter who skip the programme. If session attendance is a goal, adding sessions will not reach them. Finding out why they came is the place to start, and the post-event follow-up is a good moment to ask.
When the alternative wins: a large medical congress may genuinely need breadth, because different specialties come for different sessions; see CME credits at medical conferences. There, more sessions serve more audiences rather than the same audience more. The diminishing return applies per person, not per specialty.
Where Aurentex sits
Every figure above came from session check-in data, and your own will beat any benchmark. On Premium, Aurentex records session-by-session attendance against each delegate's registration, so after one event you know your own sessions per person, your own reach and your own quiet slots. Your first event is free.
Common questions
How many sessions does an attendee attend?
About 5.5 at the median event, between 3.3 and 8.0 for the middle half.1
Does adding sessions increase attendance?
Less than proportionally: eight times the programme for about 2.5 times the attendance per person.1
Do time of day and session length matter?
Very little. A modest late-afternoon dip, and no measurable effect of length.1
Can I size rooms from bookmarks?
Only roughly. Just 28% of sessions landed within 20% of their bookmark count.1