Methodology & definitions

The Event Bench turns public information about real B2B events into benchmarks you can plan against. This page explains where the data comes from, how a benchmark is built, and exactly what each number means.

Where the data comes from

Every figure is computed from public-source event data — the same things a diligent planner could find by hand, just gathered and structured at scale:

registration and ticketing pages
sponsor and exhibitor prospectuses
published agendas, programs and session catalogs
public event and organizer websites

These records are consolidated into a continuously updated index of event editions. We do not use private EventMobi customer data, and nothing you enter into the form is added to the benchmark set.

What's included — and what isn't

Included: real business-to-business events — conferences, trade shows, summits and professional meetings — across in-person, virtual and hybrid formats, spanning the industry verticals listed on the home page.

Not included:

Consumer / B2C events are out of scope.
Identifying details — event name, organizer, venue and source — are never exposed publicly. Benchmarks are always aggregated across a peer set, and even the data explorer shows only location and classification, never the event itself.
A handful of internal-only signals we capture are held entirely on the backend and are never served to the public site.

Series vs. editions (how we count)

A series is the recurring event brand — an annual conference that returns year after year. An edition is one specific staging of it: a single year, with that year's pricing, program and sponsor structure.

We benchmark editions, not series. A conference that has run for five years contributes up to five editions — one per year we've captured — each as its own comparable data point. The series record exists only to carry attributes shared across its editions (the organizer's type and industry classification); it is never counted as an event on its own.

This is deliberate: collapsing every year of a long-running event into a single row would bury price and program changes over time and quietly over-weight one-off events. Counting editions keeps each year comparable and keeps year-over-year movement visible. The "events indexed" figure you see is a count of editions.

Pricing, normalized to USD

Ticket prices and sponsor-tier costs are reported in different currencies at the source. Before anything is compared, every monetary value is normalized to US dollars using approximate reference exchange rates (today, USD and CAD — roughly one in ten priced events is in CAD). A price in a currency we can't yet convert is excluded from that figure rather than mixed in at face value, so a CAD $1,000 ticket is never averaged as if it were USD $1,000.

These rates are approximate and refreshed periodically. (On the roadmap: storing a normalized price at the moment each record is captured.)

How a benchmark is computed

When you submit the form, the tool assembles a peer set of editions matching your inputs:

format (in-person / virtual / hybrid)
region
size — a band of roughly half to double your expected attendance (±2×)
organizer type and industry, where the data supports them

Over that peer set it computes medians, percentile ranges and prevalence rates for each metric. We lead with medians and percentile bands rather than simple averages, because event metrics are skewed — a few very large events would otherwise pull every average upward.

When matches are thin

If fewer than 20 close peers match, the tool progressively widens the lens to find enough signal — first dropping the industry filter, then organizer type — and tells you on the results page exactly what it changed. Widened results, and any genuinely small sample, are flagged inline. When you see that flag, read the medians rather than exact values, and treat the figures as directional.

Missing data, zeros and confidence

Public records are uneven — not every event publishes every detail. We handle that explicitly rather than letting gaps distort the numbers:

A reported 0 for a count (exhibitors, sessions, tiers, speakers) is treated as "not found," not a true zero, and excluded from medians and distributions. Otherwise unpublished data would drag every benchmark toward zero.
$0 ticket entries — comp, expo, student or spouse passes — are excluded from price ranges so they don't pull the floor of the range down.
Each section reports its own sample size (for example, "based on the events with exhibitor data"), because a metric is only as strong as the subset of editions that actually published it. A figure is never implied to rest on more events than it really does.

A note on industry classification

Industry is inferred from the organizer's NAICS classification, which describes the organizing entity, not the event's subject. Many organizers are associations or trade-show producers whose code carries no topic signal (a healthcare conference run by a bank can be coded as banking). As a result, some industry filters match few editions and fall back to a broader peer set — and we flag it when that happens.

Refresh cadence

The index is updated continuously, on a rolling basis, as new editions are captured and existing records are revised. The "events indexed" count and the editions-span years shown across the site are live — read from the data at page load, never hardcoded. (Headline counts are rounded down to the nearest hundred.)

Metric definitions

What each number on your report means.

Metric	Definition
Full price	The standard paid admission for one attendee. We use the lowest paid ticket as the event's entry price; $0 comp / expo / student passes are excluded.
Median vs. average	The median is the middle value; the average is the mean. We headline the median because a few mega-events pull averages upward.
Percentile	The value below which a given share of peers fall. The 25th percentile (p25) means a quarter of peers are below it; the median is the 50th.
Trimmed extremes / "middle 80%"	The headline range spans the 10th to 90th percentile of peers, so one unusually cheap or expensive outlier doesn't define the band. The raw minimum and maximum are set aside.
Ticket tier	A ticket option an attendee can buy — early-bird, standard, VIP, virtual, and so on. "Typical tier count" is the median number of options peers offer.
Sponsor tier	A sponsorship package an organizer sells to sponsors (Gold, Silver, etc.) — a different thing from a ticket tier, which attendees buy. Equivalent tier names ("Gold," "Gold Sponsor," "Gold Level") are merged into one canonical tier so price and prevalence aren't split across spellings.
Speaker count	The number of speakers / presenters on the program. Median across peers.
Tracks	Parallel content streams running at the same time (e.g. three concurrent tracks).
Sessions	The total number of individual sessions / talks on the agenda. Median across peers.
CEU / accreditation	Whether the event offers continuing-education or professional-accreditation credit, reported as the share of peers that do.
Member pricing	Whether a discounted rate is offered to members, reported as prevalence and — where we can compute it — the typical (median) discount versus the retail rate.
Discount prevalence	Share of peers with published pricing where we could identify an early-bird, member, or student rate. The denominator is peers whose pricing we captured, not the whole peer set — an event whose pricing page we never read is unknown, not "offers no student rate." Depth is the median saving, and the baselines differ: early-bird and member compare like-for-like tiers (early vs regular, member vs non-member), while student compares the student rate against the event's top standard rate — which is why student depth reads far deeper than the other two. Early-bird prevalence also counts events where we found a cutoff date but no measurable price gap, so prevalence and depth rest on different sample sizes.
Attendee-to-exhibitor ratio	Median attendees ÷ median exhibitors across peers, shown as "N : 1" — one exhibitor for every N attendees. Computed median-over-median so mega-event outliers don't distort it.