Counting exhibiting companies when co exhibitors and pavilions blur the line

The marketing director wanted to put "over 1,500 exhibitors" on next year's sales brochure. The commercial director said the show had 1,180 exhibitors and had the contract report to prove it. Both were reading their own system correctly.

The gap was 360 companies, roughly a fifth of the floor, and it had been there for years without anybody reconciling it, because the two figures lived in two systems and were never printed on the same page.

That is the normal state of affairs. Attendance gets argued about because it is the headline, and the ground is well trodden in attendee analytics: the four slots of a metric definition and the choice of unique key both have settled answers. Exhibitor count gets published from whichever system the author happened to have open.

Which population are you counting when you count exhibiting companies?

There are three counts on any show floor and each of them is legitimate.

The contracted account is the entity that signed for space and receives the invoice. It is a commercial fact with a document behind it, it reconciles to revenue, and it is the population your sales team manages.

The exhibiting brand is the company presenting product on the floor with its own people. That includes co-exhibitors sharing somebody else's stand and companies exhibiting inside a pavilion under a single contract, and it excludes the pavilion organiser who signed the contract and sells nothing.

The directory listing is whoever appears in the show catalogue and the online exhibitor list. It is the count a visitor experiences, it is the one a competitor will scrape, and it is usually the largest.

The three differ by populations, so no amount of care in the query brings them together. They answer different questions.

Somebody has already written the definitions down

The distinctions have formal definitions in the German market, published by FKM, the testing body for trade fair figures managed by AUMA, the Association of the German Trade Fair Industry, and revised for its 2024 certification standards.

FKM defines an exhibitor as an entity granted approval for an exhibitor stand space by the organiser, which presents or distributes products, services or rights there with personnel present. A co-exhibitor is an exhibitor who, with the organiser's consent, presents products, services or rights with personnel on the stand of the main exhibitor. An additionally represented company displays goods on an exhibitor's stand without deploying its own staff, is not counted as an exhibitor, and may appear in a directory if it is marked as such.

Three things in those definitions are worth stealing regardless of which market you are in. Approval by the organiser, so an uninvited presence is not an exhibitor. Personnel present, which is the test that separates a co-exhibitor from an additionally represented company. And the explicit permission for the directory to be larger than the exhibitor count, as long as the extra entries are marked.

Whether a national body certifies your figure is a different question with its own rules, and the P cluster covers audit and verification properly. Exhibition terminology standards more broadly, and the words organisers keep using differently, are their own subject. What I am taking from FKM here is the vocabulary, because having three named populations is what lets you publish three numbers without looking evasive.

The bridge, on one edition

A show with 1,180 contracted accounts, of which nine are pavilion organisers: national and regional groups that contracted floor space on behalf of their member companies and exhibit nothing themselves.

Start at 1,180 and work to the directory. Take out the nine pavilion organisers, since they present no product, leaving 1,171. Add the 212 companies exhibiting inside those pavilions, each with its own staff and its own listing, giving 1,383. Add 157 co-exhibitors sharing commercial stands with their own staff, giving 1,540.

The bridge is minus 9, plus 212, plus 157, which is plus 360, and 1,180 plus 360 is 1,540. It has the same shape as the numbered ledger that runs from registrations to attendance, applied to the other population.

The show catalogue actually carries 1,605 entries. The extra 65 are additionally represented companies, whose products appear on somebody else's stand without their own staff, and they are marked in the directory and excluded from the exhibiting count. Marking them is what keeps 1,540 defensible.

Now the third count. Resolve those 1,540 listings to parent companies and 230 of them disappear into a shared owner: 190 pairs of subsidiaries under one parent, contributing 190 collapses, and 20 groups of three, contributing 40. 190 plus 40 is 230, and 1,540 minus 230 is 1,310 unique parent companies.

Three published counts from one floor: 1,180 contracted accounts, 1,540 exhibiting companies, 1,310 parent companies. The largest is 30.5 per cent above the smallest.

The year-on-year trap is worse than the level

Publishing one of three numbers is a disclosure problem. Comparing across editions with an unstable mix is a measurement problem, and it is the one that misleads people who are being careful.

The prior edition had 1,145 contracted accounts and 1,398 directory-eligible exhibiting companies, with six pavilions holding 148 participants between them. This edition has nine pavilions holding 212.

On contracted accounts the show grew 35, which on 1,145 is 3.1 per cent. On exhibiting companies it grew 142, which on 1,398 is 10.2 per cent.

Almost half the difference between those two growth rates is one commercial decision: three additional pavilion deals, worth 64 extra participating companies. A reader looking at 10.2 per cent concludes the sales team had an excellent year. A reader looking at 3.1 per cent concludes it was flat. The truth is that the show sold three pavilion contracts and roughly held its direct book, which neither headline says.

The CEIR Index, published through IAEE, carries number of exhibiting companies as one of its four components, alongside net square feet of exhibit space sold, professional attendance, and gross revenue. Any show contributing to a benchmark like that, or comparing itself against one, is making an implicit claim about which of the three counts it is reporting, and the pavilion mix is exactly the thing that makes two shows in the same sector incomparable on this metric.

Holding the same count across five editions is the job of the report specification, and the count is worth naming there explicitly, with a changelog row the year the choice moves.

Pavilions are the case that breaks any single rule

Every clean counting rule survives until a country pavilion arrives, and then it has to be patched.

The trouble is that a pavilion is one commercial relationship and many exhibiting relationships, and the ratio between the two is set by somebody outside your organisation. A trade promotion agency decides whether to bring 12 companies or 40, in a year when its budget moved for reasons unconnected with your show. Your exhibiting company count then moves by 28 and your contract count does not move at all.

So handle pavilions explicitly rather than letting them wash through the totals. Three practices are worth having.

Flag the pavilion organiser as a distinct account type in the contract system, so a query can separate the nine from the 1,171 without anybody remembering which nine they were. Most organisers do not have this field and adding it is a fifteen-minute change that pays for itself the first time somebody asks.

Require a participant list as a contractual deliverable, with a company name and a country for each, submitted before the directory closes. Where that list is missing, the pavilion contributes one exhibiting company, not forty, and say so in the derivation. Counting forty companies you cannot name is inventing exhibitors.

Report pavilion participants as a named line in the exhibitor derivation section, with the prior edition next to it. That single line, 212 against 148, explains most of the difference between your two growth rates before anybody has to ask.

Pavilion commercial terms are a separate subject, covering what happens when one buyer takes forty stands at once. What matters for the report is only that the count is decomposable.

Which of the three counts should you publish?

I would publish the contracted account count as the reported figure, with the exhibiting company count and the parent company count immediately beneath it in the derivation section.

The reason is that the contracted account is the only one of the three that reconciles to something else. It ties to revenue, it ties to the space book, and it cannot quietly drift, because a change in the count implies a change in the invoice ledger and somebody in finance will notice. The exhibiting company count depends on how diligently co-exhibitor declarations were collected, which varies with how busy the operations team was in the last three weeks before the show.

That is an argument about auditability rather than about which number is most true. If your question is how many companies a visitor could have met, 1,540 is the better answer and I would say so on the page.

What I would not do is publish the largest number without saying which population it counts. The exhibitor satisfaction work in the 2025 Channel Insights Report, produced by Explori with UFI and supported by SISO across more than 3,000 events surveyed between 2017 and 2025, tracks measures including likelihood of return at exhibitor level. A show that surveys contracted accounts and reports response rates against a directory-based denominator will understate its response rate by about a fifth on the numbers above, for no reason other than a mismatched population.

Where this stops

The parent company count is the weakest of the three and I would treat it with suspicion.

Resolving 1,540 company names to corporate parents means matching text that exhibitors typed into a contract form, where the same firm appears as three spellings and a divisional name. The 230 collapses in the worked example above assume a resolution process somebody built and tuned. Without one, a parent count is a name-matching artefact and will move between editions because a sales admin typed something differently.

Co-exhibitor declarations are the other soft spot. Most organisers charge a co-exhibitor fee, which gives exhibitors an incentive not to declare, so the exhibiting company count is systematically low by an amount nobody measures. A floor walk on day one counting stands with more than one brand on the fascia will size it, roughly, and it is the only cheap check I know.

Pull your contract report and your directory export for the last edition, count the rows in each, and list the pavilion organisers by name with the number of participants behind each. Those two counts and that list are ninety per cent of this bridge, and most organisers can assemble them in a morning.