Cost to attract an attendee and the three ways teams get it wrong

Two people from the same organiser present the same show at the same portfolio review. One says the cost to attract an attendee was 26. The other says 72.

Neither has made an arithmetic error. They spent an identical amount of money on an identical audience and drew the boundaries in different places, and the boundaries are worth nearly three times the number itself.

That spread is the reason cost to attract an attendee gets quoted confidently inside a company and never compared meaningfully between two. It is fixable, and the fix is two sentences of definition rather than any new instrumentation.

How do you compute cost to attract an attendee?

Total audience acquisition spend of 1.34 million, divided by professional attendance of 27,900, is 48.03. Call it 48 per attendee.

The numerator breaks down as paid media of 720,000, email platform and data costs of 145,000, telemarketing of 210,000, agency fees of 165,000, and print and trade media of 100,000. Those five lines total 1,340,000.

The denominator is professional attendance, meaning the audience left after exhibitor staff, media, organiser and contractor badges, speakers and accompanying persons have been taken out. Working out which badges those are is the whole of excluding staff badges from attendance in D12, and the important thing here is that the figure is the audience an exhibitor paid to meet.

That population has an auditable definition behind it. UFI's Calculation Standards and Definitions, which is Annex 2 of its auditing rules for UFI Approved Events, keeps staff of exhibiting companies, speakers and media representatives out of the visitor count, and then defines total attendance as unique visitors with exactly those three groups added back on top. Dividing your spend by total attendance and calling the result a cost per attendee is defensible only if you say that is what you did, and the two denominators differ by a fifth or more at most shows.

That is the honest version. Everything below is a way of arriving at a different number from the same file.

The first mistake: registrations in the denominator

Registrations are the number every audience team has to hand, they arrive earlier, and they are larger. All three properties make them the wrong denominator.

The same show had 38,600 registrations. Dividing 1,340,000 by 38,600 gives 34.72, call it 35. The same spend, the same show, the same audience, and a figure 28 per cent lower than the honest one, or put the other way, the honest figure is 38 per cent higher than the flattering one.

The gap is no-shows, and no-show rates on free trade shows are large and volatile. A show whose no-show rate moves from 28 to 34 per cent between editions will report an improving cost per registration and a worsening cost per attendee from identical spend and identical results. One of those series is measuring your marketing and the other is measuring your no-show rate.

CEIR's Cost to Attract Attendees study, aimed specifically at business to business exhibition organisers and reported by Exhibit City News on 7 May 2014, gives median spending both overall and on a per verified attendee basis, broken out by gross revenue, verified attendance, net square footage of paid space, geographic scope of attendance, whether the exhibition rotates location and the type of organiser. The choice of a verified attendance denominator is the study's, and it is the right one. If you want to sit anywhere near a published benchmark, use the same base.

Cost per registration is a genuinely useful operating measure during the campaign, when attendance does not exist yet. It belongs in the pacing report rather than the post-show report, and the blended cost per registration that supports channel decisions is B39's.

The second mistake: a numerator that only counts media

Ask a paid media manager what the show cost to fill and you will often get the media figure, because that is the budget they hold.

Paid media alone is 720,000. Over 27,900 that is 25.81, call it 26. That is the low number from the opening scene, and it is 46 per cent of the true cost.

The lines that go missing are always the same ones. Telemarketing, at 210,000 here, is the largest and it disappears because it sits in a different team's budget. Agency retainers get treated as overhead. Email platform and data costs get counted as technology. In-house salaries for the audience marketing team are almost universally excluded, which is defensible in a direct-cost frame as long as it is stated, and indefensible when the number is compared to somebody who included them.

The test for whether a line belongs in the numerator is the same test that works on the cost side of a show P and L. Would this cost exist if you stopped trying to fill this show? Telemarketing, no. Agency fee, no. The group's marketing automation licence, mostly yes, so it stays out.

Write the five or six lines down once, in the report, as a list with amounts. It costs four lines of a table and it makes the number reproducible by somebody who was not in the meeting.

There is a timing question buried in the numerator that causes its own quiet drift. Spend for an October show starts in the previous November and finishes the week before doors open, which straddles two financial years for most organisers. If your finance system reports audience marketing by fiscal period and your show report reports it by edition, the two will never agree, and the difference is whichever way the campaign window happened to fall.

Match spend to the edition it was buying. That means a campaign calendar with an edition tag on every purchase order, agreed with finance once, and it is worth the argument. A show that quietly switches between edition-basis and period-basis reporting between years will produce a cost per attendee series with a step in it that nobody can explain, and somebody will spend a fortnight looking for a marketing cause that does not exist.

The third mistake: mismatching the numerator and denominator populations

This is the subtle one and it produces the largest error of the three.

Of the 27,900 professional attendees, 9,400 arrived on exhibitor guest codes. Your marketing spend did not find those people, their suppliers did. If you keep them in the denominator you are crediting your budget with an audience you did not buy.

Take them out and the marketing-attributable denominator is 18,500. 1,340,000 divided by 18,500 is 72.43, call it 72. That is the high number from the opening scene, and it is a legitimate calculation of a different quantity: the cost of the attendees your own programme produced.

Four numbers, then, from one show: 26, 35, 48 and 72. Each is defensible, each answers a different question, and the only indefensible thing is quoting one without saying which.

My preference is to publish 48 as the headline, because total spend over total professional attendance is the figure that reconciles to the show P and L and is hardest to game. Then publish 72 alongside it as the marketing-attributable figure, because that is the one that should inform whether the budget is working. The two together also make the exhibitor guest code programme visible as the acquisition channel it actually is, which is usually the cheapest audience the show has and the least managed.

What you cannot do is mix the populations. Excluding guest code attendees from the denominator while leaving the cost of promoting the guest code programme to exhibitors in the numerator gives a number that is wrong in a direction nobody notices.

What is the figure actually for?

Cost to attract an attendee is not a target to be minimised, and treating it as one produces predictable damage.

The cheapest attendee is the one you did nothing to reach, so any team scored on this metric alone will cut the channels that find new people and keep the ones that harvest the base. Cost per attendee falls, the audience ages, and three editions later the first-time share has collapsed.

The figure earns its place in three other uses. It sizes next year's budget: if you need 29,000 professional attendance and your programme-attributable cost runs at 72, and you expect a third of the audience from guest codes, the arithmetic gives you a budget rather than a negotiation. It benchmarks across a portfolio, where two shows in the same vertical at 48 and 91 is a question worth asking. And it feeds the event ROI calculation, where 1.34 million of audience acquisition sits inside direct show cost alongside venue and operations, which is D17's territory.

For the portfolio comparison to mean anything, every show in it has to use the same boundary. That is a policy decision made once, centrally, and written into the reporting definition. It is not an analytical problem, and it is the reason most portfolio comparisons of this metric are worthless.

Where this stops

The denominator rests on a verified attendance figure, and a show whose badge scan coverage is poor does not have one.

If 12 per cent of your attendees never crossed a scanner because a side door was open on day one, your denominator is 12 per cent low and your cost per attendee is 14 per cent high. That error is invisible in the ratio and it will be read as a marketing efficiency problem. The scan coverage figure belongs next to the cost figure for exactly this reason, and it is the point where this metric depends on the rest of your attendee analytics being sound.

The second limit is attribution. Splitting the audience into programme-attributable and everything else assumes you can tell which is which, and last-touch registration source data is a poor instrument for that. Somebody who saw a trade press advertisement in June, remembered the show in September and registered through an exhibitor code is recorded entirely against the exhibitor. The 18,500 denominator is therefore too small and the 72 too high, by an unknown amount.

The third is that the metric has nothing to say about who those attendees were. 48 per attendee for an audience of specifiers with budget is a bargain and 48 for an audience of students is not, and the ratio is identical. A version weighted by qualification is more useful and considerably harder to defend, since somebody has to decide what qualified means and that decision moves the number more than any of the boundary choices above.

Take last edition's audience acquisition spend, list every line that would disappear if the show did not run, and divide the total by your professional attendance figure rather than your registration count. Then do it again excluding guest code attendees from the denominator. The distance between those two numbers is how much of your audience your own budget is actually finding.