Verified versus registered attendance and the gap between those two numbers

There are two slides in the post-show pack. One says 18,000, taken from the registration platform on the morning doors opened. One says 14,120, taken from the audit certificate that arrived in April. Both are correct. The argument about verified versus registered attendance then runs for forty minutes, ends without a decision, and repeats next year, because nobody wrote down which base the show publishes.

The gap is worth understanding structurally, since it is made of two entirely different problems that happen to arrive as one number.

What counts as a registration and what counts as verified?

A registration is a completed form. It exists the moment somebody presses submit, it survives cancellation unless you remove it, and it has no relationship with the show floor.

Verified attendance is industry vernacular rather than a term of art, and it is worth being precise about what sits behind it. UFI's Approved Event programme requires independently verified figures, and what its rules actually define are visits and visitors. UFI's Calculation Standards and Definitions, Annex 2 to the auditing rules dated June 2021, define a visit as the action of a person who on an official open day and hour enters the event with an access document and who is representative of the market audience expected by the organisers and exhibitors, and define a visitor as a person attending the exhibition, counted once for the entire duration regardless of how many visits they made.

FKM, the German association that has published verified exhibitor and visitor data since 1966, lands in the same place from a different document, defining an exhibition visitor as "a person attending a trade fair/exhibition during the hours in which it is officially open to exhibition visitors, who wishes to obtain information and/or make contact with exhibitors" and applying the same rule that the person is counted once for the entire duration.

Both definitions turn on attendance during declared hours. Neither mentions registration. That is the whole reason the two numbers diverge.

There is no box on the certificate for registrations

Here is the fact that settles most of the internal argument, and almost nobody in the room knows it.

The UFI Standard Audit Certificate, the version UFI publishes today marked 08/21, has fields for net indoor and outdoor exhibition space, national and international exhibitors, and a visitor block laid out as visitors counted once, plus repeat visits, equalling total visits. The form carries a note that either the visitor figure or the total visits figure is compulsory. There is no field for registrations anywhere on the three pages.

So a media kit headline built on registrations is not a weaker version of an audited figure. It is a figure from a different system that no auditor has been asked to look at, sitting on a page that also carries audited figures, in the same typeface. Readers cannot tell them apart and most do not try.

Decomposing the gap into its two halves

Work an example through, because the decomposition is the useful part.

A show carries 18,000 live registrations at the moment doors open. Across the three days, 14,600 badges are collected. When the auditor joins the badge collection log to the registration export, 14,120 collections match a record that survives scrutiny.

Two subtractions, two different stories.

The first is 18,000 minus 14,600, which is 3,400 people who registered and never came. As a share of registrations that is 18.9 per cent. This is behaviour. It responds to price, to the weather, to a competing event and to how far in advance people registered, and no amount of data hygiene reduces it. The arithmetic of that population, and what an auditor does with it, belongs with registrations that never became a badge.

The second is 14,600 minus 14,120, which is 480 people who were physically present and left evidence an auditor could not follow. As a share of collections that is 3.3 per cent. This is a data problem, it is yours, and it is fixable. Unkeyed paper registrations at the door, a badge reprinted and logged twice, a group check-in scanned once for six people, a reader logging outside the declared opening hours.

Putting both together, 14,120 divided by 18,000 gives 78.4 per cent verified against registered. That single ratio is the number people quote, and quoting it without the split hides the fact that 3,400 of the 3,880 missing were a marketing outcome and 480 were an operations defect.

Which number should you publish?

My view is that you publish the verified figure as the headline and the registration figure nowhere near it.

The reasoning is about who reads the page. An exhibitor deciding whether to renew wants the count of people who could have walked past their stand. A registration that never arrived could not walk past anything. Publishing 18,000 to that reader is a claim about your marketing funnel dressed as a claim about your show floor, and exhibitors work this out eventually, usually in the renewal conversation you least wanted it raised in.

There is a real cost to the discipline and it is worth stating. If your competitor publishes registrations and you publish verified visitors, your show looks 22 per cent smaller than it is on a like-for-like basis, and no buyer is going to do the correction for you. The answer is a footnote naming the base in the same visual weight as the number, and a consistent base across every surface you control. It works slowly. It also makes your figure the one a buyer can check, which is the only durable position available.

Where registrations do belong is in your own funnel reporting, where the ratio between bases is a diagnostic rather than a claim.

Is the ratio worth tracking across editions?

Yes, with two conditions, and it is one of the more useful single numbers an event data team can maintain.

The first condition is a frozen definition. Registrations counted at which moment: launch, show open, or after the on-site desk closes. Cancellations in or out. Duplicate rows netted or left. Test records removed. Each of those choices moves the ratio by a point or two, and a series where the definition drifted is a series measuring your definitions.

The second condition is that you keep the decomposition rather than the single figure. A ratio that falls from 78.4 per cent to 74.1 per cent tells you nothing on its own. The same movement caused by 900 more non-arrivals is a marketing question about who you recruited and when. Caused by 700 more untraceable collections it is an operations question about the door. Those go to different people and only the split routes them correctly.

Do not benchmark the ratio against another show. A free expo where registration costs nothing will sit far below a paid congress where a delegate has already spent money, and both can be well run. The comparison worth making is against your own previous editions with the definition held still.

The four words a media kit uses interchangeably

Registrations, attendees, attendance and visitors get rotated through a single page as though they were synonyms for the same population. Three of the four have definitions and only one of those definitions is audited.

Visitors is the defined and certified term, counted once per person for the whole show.

Attendance has a definition too, and it is broader than most organisers realise. UFI's Calculation Standards and Definitions set total attendance as unique visitors plus exhibitor staff plus speakers plus media representatives. That is a legitimate figure with a written formula behind it, and it has no box on the audit certificate, so a show quoting total attendance is quoting an uncertified aggregate built partly from certified parts. Whether your own exhibitor staff belong in a published headline is a separate argument, covered in how exhibitor staff sit against visitor counts.

Attendees has no definition anywhere in the standard. It is the word people reach for when they want the largest defensible number and have not decided which one that is.

Registrations has a precise meaning inside your own platform and no meaning at all in the audited framework.

The practical fix is a short glossary at the front of the post-show pack, four lines long, naming which term the show uses for its headline and what the other three would have produced. Once that page exists, the annual argument becomes a document review instead of a debate, and anybody quoting the wrong figure in a proposal can be pointed at a line rather than contradicted in a meeting.

Where this stops

The verified figure is the more defensible number and it is not the true number, and it is worth being clear about the direction of the error.

Verification is a floor. Every failure mode listed above removes people who were genuinely there, and there is no corresponding mechanism that adds people who were not. If your access control at the north entrance was unstaffed for two hours on the busiest morning, the certified figure is lower than reality by however many people walked through, and it will be certified anyway because the auditor certifies what the evidence supports.

The second limit is that the verified base still contains a choice you made before fieldwork. A show can be certified on unique visitors or on total visits, and those two figures can differ by half again on a three-day show, so the base the certificate carries matters as much as the verification behind it. Two shows can both be honest, both be audited, and be uncomparable.

This week, take last edition's numbers and write the three-line decomposition: registrations at show open, badge collections, and collections traceable to a record. Put the two subtractions next to them with an owner's name against each. That is a fifteen-minute exercise, it usually surprises somebody, and it turns a forty-minute argument into a one-line footnote in the standards note you attach to next year's figures. The mechanics of how the third line gets tested are covered in what an audit checks.