The attendance denominator for sponsorship metrics decides every number above it

Two people from the same organiser quote two different audience numbers to the same sponsor in the same week. Sales says nine thousand. The post-show report says six thousand four hundred. Neither of them is lying and neither of them has looked at what the other one used, and the sponsor now has a spreadsheet with both figures in it and a question about which one the rate card was built on.

The attendance denominator for sponsorship metrics is the single input with the widest reach into every number you publish, because it sits underneath all of them. Change it and every CPM, every cost per reach point and every value claim in the fulfilment report moves in lockstep, without a single impression estimate changing.

Four numbers from one show

A mid-sized three day B2B show, and four defensible ways to say how many people came.

Registrations: 9,000. Everybody who completed a form, including the ones who never turned up. Verified attendance: 7,300. Everybody who collected a badge, so the no-shows are stripped out. Unique badge-scanned on the show floor: 6,400. Everybody whose badge was read by something on the floor at least once, which drops the conference-only delegates and the people who spent two days in the hotel bar. Average daily presence: 4,200. The number of distinct people in the building on a typical day, which is the figure that describes how crowded the aisles actually felt.

Every one of those four is a real count of a real thing. The spread from the largest to the smallest is a factor of 2.14, and the spread between the two most commonly quoted, registrations and unique scanned, is 40.6 per cent.

What does the choice do to a CPM?

Take one asset. A hall banner package at 28,000 dollars, with a stated assumption of twenty exposures per person across the show.

Against 9,000 registrations, the impression estimate is 180,000 and the CPM is 28,000 divided by 180, or 155.56. Against 7,300 verified attendees, 146,000 impressions and a CPM of 191.78. Against 6,400 unique scanned, 128,000 impressions and a CPM of 218.75. Against 4,200 daily present, 84,000 impressions and a CPM of 333.33.

One asset, one price, one exposure assumption, four CPMs from 156 to 333. Nothing about the banner changed. The only thing that moved was the number underneath it, and the largest and smallest differ by more than the entire premium you would charge for exclusivity.

This is why a sponsor who benchmarks your card against another organiser's is often comparing nothing at all. If they quote registrations and you quote unique scanned, your inventory looks 41 per cent more expensive per thousand for identical delivery.

A real audit certificate, read properly

The clearest published illustration of why the base matters comes from CES. The CES 2026 Attendance Audit Summary, drawing on the Alliance for Audited Media's CES 2026 Event Audit Report, gives total verified attendance of 148,392 for the January 2026 edition.

That total is made of three segments, and the summary prints all three. Industry attendees, 86,679. Exhibitor personnel, 54,676. Media, 7,037. So 36.8 per cent of the headline attendance figure is other exhibitors' staff, and 4.7 per cent is press.

Now consider a sponsor selling enterprise software to buyers. Priced against 148,392, a 250,000 dollar activation looks like 1.69 dollars per attendee. Priced against the 86,679 industry attendees, it is 2.88 dollars, which is 71 per cent higher. The activation is identical. The audience the sponsor actually wants is a bit under three fifths of the headline.

The point is not that CES is doing anything wrong. It is doing the opposite: it publishes the split so a buyer can pick their own denominator, which is exactly the behaviour a rate card should support. Most shows publish one number and let the buyer assume.

Which denominator should a sponsorship card use?

I would use unique verified attendance for on-site assets and treat everything else as an exception that has to be argued.

The reasoning is a hard physical constraint. No banner, lanyard, aisle sign or lounge can reach a person who did not enter the building, so attendance is a ceiling on reach for that whole category, and pricing above the ceiling is pricing against people who cannot see the thing. Registrations fail this test by definition, since a registration that never converted to a badge is a person your signage never met.

Digital assets are the exception and they need their own base. A sponsored email goes to a database that includes lapsed attendees, prospects and people in other countries, and its reachable audience is a property of the list rather than the show. Pricing an email against show attendance understates its reach badly and pricing on-site signage against the database overstates it worse. Two bases, each stated, each attached to the asset types it governs.

Average daily presence has one legitimate use, which is capacity-limited inventory. A charging lounge with 60 seats or a sponsored shuttle with a fixed timetable is bounded by how many people are around at once, and the daily figure describes that better than the cumulative one.

Writing the rule down once

The failure mode is not choosing badly. It is choosing differently in different rooms.

The fix is a definitions note that lives with the rate card and appears in the fulfilment report template. Three or four lines. Which figure is the on-site base, which is the digital base, how each is derived, and the date after the show when each is final. Then every CPM, every impressions estimate and every renewal comparison uses those two numbers and nothing else, and last year's report is comparable to this one because the definitions did not drift.

UFI's rules for the statistics of its Approved Events make the same demand of audited figures. Visits, meaning visitors plus their repeat visits, may be reported alongside visitors, but the standard audit certificate has to state clearly which is which and by what method it was calculated. That is the whole discipline in one requirement: report whichever figure you like, and never let the reader guess which one they are holding.

There is a demand-side reason to care as well. The ANA and MASB survey published in July 2018 as Improving Sponsorship Accountability Metrics found that only 30 per cent of sponsors audit or verify the metrics they receive from the property, against 42 per cent who do not. Most of your sponsors are taking your denominator on trust, which is an argument for getting it right rather than an argument for it not mattering. The 30 per cent who do check are the ones who will notice the year the base quietly changed, and they are disproportionately your largest accounts.

Where this stops

The denominator can be chosen carefully and still be wrong, because it is only as good as the collection underneath it.

Badge scan coverage is the usual culprit. If your floor has scanners at four of nine entrances, unique badge-scanned attendance is an undercount of unknown size, and switching to it from registrations replaces a figure that is too high with a figure that is too low. That is progress in honesty and not necessarily in accuracy. The right move is to state the coverage next to the figure, so a reader knows 6,400 came from four gates rather than the full perimeter.

The second limit is timing. Registrations are final about six weeks after the show once the late data reconciles, verified attendance settles sooner, and unique scanned depends on when the last scanner file is loaded. A fulfilment report written on the Friday after close is using provisional numbers for at least one of them, and if the sponsor report goes out early and the audit certificate arrives later with a different figure, you have created the discrepancy yourself. Pick the base, then pick the date you will freeze it, and put both in the definitions note.

How that figure gets audited, and what separates a verified count from a registered one in the first place, belongs to the verified against registered distinction and its neighbours. What sits on top of it, the per-asset ranking in one CPM column across the card and the overlap arithmetic in splitting gross impressions from net reach, all inherit whatever you choose here. It is the same exhibitor analytics chain, and the base is the first link.

Pull the last sponsorship rate card and the last post-show sponsor report side by side and find the attendance figure each was built on. If they differ, you have already sent a sponsor two incompatible numbers, and the fix is one line in a definitions note before the next card is printed.