Trade show marketing measurement: denominators, sample size and your own ranges

What to take away

Name the denominator before the show openseligible attendee, assessed account, booth zone entry, qualified opportunity.
Report every rate with its sample size. A 38 percent fit rate from 120 assessments carries about plus or minus 9 points at 95 percent confidence.
Treat published ranges as questions to test, never as quotas.
Renew on mature incremental value against full cost, not on badge scans.
Date every definition so next year's series stays comparable.

Denominator first

Registrations are not attendees.
Attendees are not suitable buyers.
Aisle traffic is not booth engagement.
Scans are not qualified opportunities.
Attributed revenue is not incremental contribution.

Each pair has its own denominator. Mixing them produces a number nobody can defend in a budget review. The planning sequence that the counting rule has to fit inside is set out in trade show marketing.

Write down three things before the doors open: who counts as an eligible person, which entries count once, and what counts as booth contact.

The measurement ladder

Measure	Denominator	Typical range	Main limit
Show attendance	Eligible person, one entry rule	Organizer headline less 15 to 30 percent	Organizer definition, duplicates
Audience fit	Assessed attendees or accounts	25 to 45 percent match the target profile	Sampling, missing roles
Booth approach	Observable passersby or zone entries	10 to 25 percent of eligible attendees	Floor position, counting method
Useful exchange	Approaches or conversations	30 to 50 percent of approaches	Quality definition
Requested follow-up	Useful exchanges	60 to 80 percent	Permission, record accuracy
Qualified outcome	Eligible contacts or accounts	20 to 40 percent of follow-ups	Sales lag, prior pipeline
Incremental return	Matched comparison, full cost	Depends on deal cycle	Selection, spillover, uncertainty

The middle column is illustrative, built from your own past events and split by event type, audience, geography and deal cycle. No single body publishes booth-level trade show performance norms, so treat any range as a question rather than a target.

Before you fix a ladder, settle what the show is for. The trade show marketing questions worth asking before anything else covers that ground.

NIST's page on measurement uncertainty makes one habit worth copying: publish the value with its uncertainty attached, not the value alone. Borrow that habit and leave the rest of the metrology vocabulary out of the report.

Example: one regional show, worked end to end

The figures below are an illustrative worked example, not a real event.

One regional show, worked end to end

The organizer reports 4,200 registrations. Door scans return 3,100 unique badges. After removing exhibitor passes and duplicate entries, eligible attendees sit between 2,900 and 3,300. Denominator: eligible attendee.
Staff assess 120 scanned badges against the target account profile. 46 match, so fit is 38 percent, with a 95 percent interval of about 29 to 47 percent. Denominator: assessed attendee.
Zone counters log 610 booth entries. Conversations reach 240. Denominator: booth zone entry, then conversation.
Follow-up requests total 180. Ninety days out, 62 are qualified opportunities. Denominator: requested follow-up.
Full cost is 48,000 dollars for space, build, travel, staff time and follow-up hours. Attributed closed revenue at nine months is 74,000 dollars. Cost per qualified opportunity is 774 dollars.

A benchmark you cannot recompute from its source is a rumor with a decimal point.

Sample size before you quote a rate

To report a proportion within plus or minus 5 points at 95 percent confidence, the standard calculation needs about 385 completed assessments. The arithmetic is 1.96 squared times 0.25, divided by 0.05 squared.

At 120 assessments the interval is near plus or minus 9 points.
Below roughly 1,000 attendees, a finite population correction narrows the interval.
Under 30 units, quote the count, not a rate.

Give every printed QR code a unique destination built with Google's Campaign URL Builder, so scans land in the same dataset as the booth records.

Reference ranges and their sources

Cost per square foot has a public reference point. Wikipedia's article on trade shows describes booth space rental and the factors that move per-square-foot pricing, which frames a cost question without setting your budget.

For everything else, build reference classes from your own history and split them by event type, audience, geography, participation format and deal cycle.

Check the rules that touch what you hand out. Giveaways for children carry testing and labeling duties under eCFR Part 1107. Canadian promotional products sit under the Canada Consumer Product Safety Act and Health Canada guidance, and commercial print in Quebec follows the Charter of the French Language.

Benchmarks in a neighbouring channel are built the same way. Direct mail marketing benchmarks shows the arithmetic for cost per response, where a missing denominator ruins the figure just as fast.

The record beside the decision

Publish the population, period, inclusion rule, calculation, currency, maturity window and uncertainty for every figure. Recompute it from the cited source and note any mismatch. Record who owns each count in a trade show marketing checklist so a reviewer can reproduce the reasoning without asking you.

Set a review date and name the change that would trigger an earlier check.

Common questions

What is a good scan-to-lead rate?

There is no universal figure. Scanning rules, audience, qualification and permission differ, and the denominator moves with them. Your own past shows are the only defensible comparison.

Can organizer attendance support a forecast?

Only as bounded context, and only after you understand how the organizer counts. It does not predict booth traffic or revenue.

Which benchmark should guide renewal?

Mature incremental value against complete cost, with a stated window and uncertainty. Nothing else survives a second look.

How many past shows do I need before a range means anything?

Three is a starting point, five is better. Label anything built on fewer than three as a single observation.