Event metrics worth tracking

Event dashboards are usually full and rarely useful. The reason is that most of what they contain rises when you do more of something rather than when the event gets better, so nobody can act on any of it.

The test for a metric is the same as for a survey question: name the decision it changes. Twelve numbers pass that test at most conferences. Here they are, with how to calculate each and what it tells you.

The twelve

1. Attendance against registration

Attendees who arrived, divided by attendees who registered.

Why: it sizes everything else, and every per-head cost in your budget depends on it. A no-show rate that moves year over year is the single most useful planning number you have.

Decision it changes: catering guarantees, room allocation, and how many tickets you sell against a fixed capacity.

2. Activated attendees

People who claimed their invite and opened the app, divided by attendance.

Why: it is the honest measure of app reach, and it is the number our own billing runs on. It excludes people you imported, people who were invited, and people who registered and stayed home.

Decision it changes: whether the app is worth its budget, and whether your invite process needs work. A low number on day one is almost always an invite problem rather than an app problem.

3. Session attendance against room capacity

Per session, and per track.

Why: it is the only reliable evidence of what your audience actually wanted, and it is more honest than a survey because it is behaviour rather than opinion.

Decision it changes: next year’s track sizes and room allocation. A track that ran at 40% capacity all day is a track to shrink or reposition.

4. Personal agenda rate

Attendees who saved at least one session, divided by activated attendees.

Why: it measures intent before the event and predicts session attendance better than registration does.

Decision it changes: which sessions need a bigger room, decided before the day rather than during it.

5. Question participation

Attendees who asked or upvoted at least one question, divided by activated attendees.

Why: it measures whether your event is a room where people do things. It is the single best proxy for engagement that does not require a survey.

Decision it changes: whether to staff more Q&A moderation, and which session formats to run more of.

6. Connections per attendee, and the share who made one

Two numbers, and you need both.

Why: the average alone hides the failure mode. A high average with a low share means a few people met everybody and most met nobody, which looks like success on a dashboard and feels like failure in the room.

Decision it changes: which networking formats to keep. See the networking playbook.

7. Session rating, per session

The average of the in-session rating, with the response count next to it.

Why: rated in the room, by people who just left it. This is the only session feedback worth acting on, and it is a different instrument from the post-event survey.

Decision it changes: who you invite back.

8. Net recommendation

The recommendation question, with the free-text follow-up.

Why: one comparable headline number across years. The value is in the reasons, not the score.

Decision it changes: almost nothing on its own. Its job is to be the number you compare to last year’s, and to route you to the free text.

9. Cost per attendee

Total cost divided by attendance.

Why: it is how you know whether the event scales and what a ticket must cost. See the budget template.

Decision it changes: ticket price, sponsorship targets, and which line to cut.

10. Revenue per attendee

Total income divided by attendance.

Why: paired with the line above, it is the whole commercial picture in two numbers.

Decision it changes: whether to grow the event or improve it.

11. Repeat attendance rate

Attendees who came last year and came again, divided by last year’s attendance.

Why: the most under-used number in conference operations. It is a truer verdict on last year’s event than any survey, because it cost the attendee money to give it.

Decision it changes: whether the problem is acquisition or retention, which are entirely different problems with entirely different fixes.

12. First-time attendee share

New attendees divided by total attendance.

Why: it tells you whether the event is growing or recirculating, and it should be read against the number above. Both high is healthy growth. Both low is a shrinking event with a loyal core.

Decision it changes: where marketing spend goes.

Stop reporting these four

Total app opens. Rises when you send more notifications. Measures your notification volume.

Total notifications sent. An input, presented as an outcome.

Total sessions on the programme. Rises when you programme more, which is not the same as programming better.

Social media impressions. Not comparable across platforms, not comparable across years, and not attributable to anything you did.

Each of these appears in event reports because it is easy to get and it goes up. None of them changes a decision.

Measuring during the event

Three numbers, watched live, that let you act while it still matters:

Number	Watch it	If it is off
Arrivals against the guest list	Through the morning of day one	A gap usually means a scanner problem, not an attendance problem. Check the desk before you check the forecast
Activation, hourly on day one	Through the morning	If it stalls, the invite or the registration desk is the problem, and both are fixable that morning
Session attendance, at the start of each block	Each transition	A packed room and an empty one in the same slot is a room-swap you can still make

event67’s engagement analytics run live during the event for exactly this reason, alongside a live arrivals count on the console and at the desk. The value of a live number is that somebody can do something about it; a post-event report of the same number is a lesson for next year.

Measuring ROI honestly

Most event ROI reporting is unconvincing because it applies one formula to three different kinds of event.

A revenue event — you are selling something. Measure pipeline influenced and closed, over a window you set in advance. Set the window before the event, because setting it afterwards is how attribution becomes storytelling.

A member or community event — you are retaining people. Measure renewal rate among attendees against non-attendees. This comparison is the strongest ROI evidence available to any association and it is rarely run.

An internal event — you are changing behaviour. Measure the behaviour. If the sales kickoff was about a new methodology, measure adoption of the methodology at 30 and 90 days.

A brand or community-building event — be honest that the measurable outcomes are weak and say what you are actually optimising. Repeat attendance and recommendation are the least-bad proxies. Do not manufacture a revenue number to satisfy a board; a clearly stated qualitative case survives scrutiny better than a fabricated quantitative one.

Rules for a dashboard people use

Twelve numbers, held constant. The value of an event metric is almost entirely its trend, and a dashboard that changes composition every year never produces one.

Compare to yourself. Industry benchmarks in this category are not comparable, because the definitions differ, the audiences differ and the counting methods are usually undisclosed. Your own last event is the only fair comparison you have.

Put the denominator on the page. “62% question participation” means nothing without knowing whether that is of activated attendees or of everyone who walked in. Write the definition next to the number, once, and never change it.

Segment by one thing. First-time against returning attendees. If you split by nothing else, split by that, because the two groups experienced different events.

Write the decision next to the metric. If you cannot, delete the row. That single discipline will take most dashboards from forty numbers to twelve, which is the point.