One report says 60% of registrants attend, another says 33% — both are right, and the difference decides what you build

Two webinar benchmarks get quoted in the same slide deck and they do not agree. One puts registration-to-attendance at 60%, up from 57% the year before, with an average of 239 attendees engaging for 51 minutes against a 49-minute runtime. The other, across 19,531 webinars from 418 brands, puts it at 33%, with an average of 238 registrants producing 51 attendees and 29 minutes of watch time. The gap is not a contradiction and it is not a sampling artefact. The first counts anyone who watches, live or on demand. The second counts only the people present while the session is running. Eighty percent of webinars are made available on demand afterwards, so most of the difference is people watching later, which is an entirely different behaviour to plan for.

60% and 33% are the same audience counted twice: once including the people who watch it later, once not. Which number you use decides what you build.

1. 01

   The reconciliation is the finding, and it has a practical consequence

   Live-only attendance around a third means two-thirds of registrants are a content audience rather than an event audience. A programme built only for the live hour treats that majority as a failure, when it is the segment the on-demand library exists for. Report both numbers, always labelled, and never average them into one.

2. 02

   Series beat one-offs by five percentage points and are already a quarter of the market

   Series webinars are 26% of all webinars measured and attend at 38% against a 33% average. In education that matters more than the gap suggests, because a recurring session at a fixed time becomes a diary entry for people whose weeks are structured by a bell schedule.

3. 03

   The cheapest registration lift in the dataset is one word in the title

   Using the word webinar in the title lifts registrations by 66%. It is a plain finding and it runs against the instinct to make a session sound like an event, a workshop or a masterclass. In a market whose audience is professionally used to attending sessions, saying what it is beats making it sound special.

4. 04

   Education paid search is the second most expensive of the twenty-three industries measured

   Education and instruction carries a $77.48 cost per lead, behind only legal, at a $4.81 cost per click and a 13.14% conversion rate. Education leads on Meta run $28.22 at a 10.08% conversion rate. The paid route to a registration is not cheap in this category, which raises the value of the list an existing programme already owns.

5. 05

   The numbers this market most wants are the ones nobody publishes

   There is no credible published cost per lead for K-12 district sales, no published sales-cycle length into districts, and no education-specific webinar benchmark from any platform. The closest institutional figure is a 2022 report putting median cost per enrolled student at $2,795 at four-year private institutions and $494 at four-year publics — different buyers, different economics, and four years old. Cite it with its date or leave it out.

Live-only attendance **~33–48%** (strict/cross-platform low end ~33%); live + on-demand **~57–60%**; roughly half of all views happen on-demand.

The model builds one programme with two measurement frames. Sessions run as a series at a fixed cadence and a fixed time rather than as separate events, because the published attendance premium for series is the cheapest structural gain available. Titles say the word webinar. Every session is produced as two artifacts from the start: the live hour, and the on-demand asset with its own landing page, its own promotion afterwards and its own registration path, since the majority of the audience in the live-only data arrives later. Reporting carries live attendance and total attendance as separate lines with the definition attached to each. Paid promotion fills the first sessions of a series and is then reduced, because at a $77.48 published cost per lead the owned list is the cheaper channel once it exists.

1. 1

   Week 1

   Definition sheet

   Live attendance and total attendance defined separately, with the reporting rule written down

   Owner: OmniFlow

2. 2

   Weeks 1–3

   Series design

   Fixed cadence, fixed time, titles that say plainly what the session is

   Owner: OmniFlow + client

3. 3

   Weeks 2–6

   On-demand build

   Every session produced as a live event and as an on-demand asset with its own page

   Owner: OmniFlow

4. 4

   Week 3 onward

   Owned-list promotion

   Segmented invitations from the existing list as the primary channel

   Owner: OmniFlow

5. 5

   Weeks 3–10

   Paid fill

   Search and Meta used to seed early sessions, reduced as the list grows

   Owner: OmniFlow

6. 6

   Monthly

   Reporting

   Live and total attendance reported as separate lines, never averaged

   Owner: OmniFlow

7. 7

   Month 6

   Review

   Cost per registration by channel, attendance by position in the series

   Owner: OmniFlow + client

The model reports live attendance and total attendance as two separate numbers with their definitions attached, because a single blended figure is what made the published benchmarks look contradictory in the first place. Cost per registration is reported by channel with the owned list separated from paid, and the published education cost per lead is used only as a ceiling test on whether paid fill is still worth running. Every published input is replaced by the client's own figures from month two. Modelled outputs are not a forecast or a guarantee of results.

Modelled. Inputs: ON24, 2026 Webinar Benchmarks Report (2025 data); Goldcast, 2025 B2B Webinar Benchmark Report (19,531 webinars, 418 B2B brands); WordStream by LocaliQ, Google Ads Benchmarks 2026 (13,474 US search campaigns, April 2025 – March 2026); WordStream by LocaliQ, Facebook Ads Benchmarks 2025 (data April 2024 – June 2025); Ruffalo Noel Levitz, 2022 Cost of Recruiting an Undergraduate Student Report (59 institutions). Modelled outputs are not a forecast or a guarantee of results. Cited: ON24 *2025 Webinar Benchmarks*; Livestorm *Webinar Benchmark Report*.

Every figure below is published research, not a client result. They are the inputs to the arithmetic above, listed so it can be checked rather than taken on trust. The bracketed number points to the full citation at the end of this page.

These are the surfaces this engagement is run and measured from, shown with representative figures built around the benchmarks cited on this page. Every account we run reports into views like these, and you keep ownership of all of them.

Each figure on this page, the system it is read from, and the definition and window it is measured over.

Not applicable — this is a modelled engagement. Its weakest input is that both webinar datasets are drawn from B2B software audiences rather than from education, and neither publisher segments by industry. A district audience whose week is fixed by a bell schedule almost certainly attends live at a different rate than a software buyer with a flexible diary, and the direction of that difference is not knowable from the published data. The definition sheet exists so that the client's own live and total figures replace both benchmarks by month two.

A modelled capability, not a client account. Figures illustrate what the model produces and are labelled as modelled wherever they appear. See all our work

Full citations for everything cited on this page, with the sample and period each study covers, so you can go and read the original.