Webinar attendance rate benchmarks, and why they disagree

What the current data says

These are the three recent datasets that say where their numbers come from. Each is published by a company that sells webinar software, so each describes that company's customers rather than webinars in general.

That's an 18-point spread for what sounds like one number. Treat a figure quoted without its sample, its year and its definition as unusable. Zoom's own statistics roundup quotes both 49% and 58% in one article, because it draws on more than one of these sources.

Why the numbers disagree

Four things move an attendance rate without anyone's webinar getting better or worse.

• The denominator. Some counts divide by everyone who registered. Others leave out people who cancelled, or people who registered after it started to get the recording.
• What counts as attending. Joining for 30 seconds usually counts. A 57% that includes replays and a 49% that doesn't can describe the same webinars, as Univid's two figures do.
• Median or mean. Univid reports medians by webinar. A few very large webinars pull an average around in a way they don't pull a median.
• Who's in the sample. A vendor's customers are not everyone. Goldcast's report covers organizations with an average of 251 registrants per webinar, and TwentyThree's averages 308. A 40-person community session behaves differently.

Be careful with any single figure attributed to "industry data" with no link. If you can't trace a statistic to a report with a sample size, don't plan around it.

So what's a good attendance rate?

A judgment, not a statistic: for a free B2B webinar with open registration, 40 to 50% live is normal. Under 30% is worth investigating. Over 60% is excellent, or it means your denominator is smaller than you think, so check that first.

Replay viewers are real audience and deserve a separate line. Univid's figures suggest replays add about 8 points on top of live attendance. Keep them apart, because your reminders and your topic affect live attendance, while your follow-up affects replays.

What about the best day?

The studies disagree here too, which is the finding. Goldcast's highest attendance rates are on Monday (47.2%) and Friday (46.9%), and it calls Tuesday the best all-round choice because of its registration volume. Univid puts Wednesday (51%) and Thursday (50%) on top. TwentyThree has Thursday at 64%.

If there were a strong universal best day, three datasets would agree. Pick the day your audience can attend, then test it. Note also that attendance rate and registration count pull apart: a day with a high rate and few registrants may leave you with a smaller room.

How to work out your own benchmark

1. Decide your definitions and write them down: who's in the denominator, and how a live attendee is counted.
2. Remove cancellations from the denominator. They told you they weren't coming; a good cancel link makes your rate more honest.
3. Count unique people, not join events. Someone who drops and rejoins is one person.
4. Record live attendance and replay viewers separately.
5. Do this for your last five events and use the median. Five isn't many, but it's your audience, and the median stops one unusual event from setting your target.

Break the rate down by source

An event-wide rate hides the most useful thing you could know. People who registered from a speaker's personal post and people who registered from a paid ad rarely turn up at the same rate, and a blended 44% tells you nothing about which to do more of.

If every registration records where it came from, you can compare attendance by way in. A channel that brings 200 registrants at 20% produces fewer people in the room than one that brings 100 at 50%. Ranking channels by attendees rather than registrations often changes the order.

In Programmable Talk each registration keeps its way in, and attendance comes back from Luma, Zoom or Eventbrite when connected, or from check-in at the door, so you can make this comparison without stitching spreadsheets together.