Event staff cost allocation when one team runs six shows a year

The April show's pack goes out with 205,000 of operations payroll sitting in its direct cost block. The operations director reads the line twice. She has twelve people. None of them were hired for the April show, all of them were on the March show three weeks earlier, and four of them were already in the June venue by the time the pack was written.

She asks finance where the 205,000 came from and gets a spreadsheet with a tab called allocation. That is event staff cost allocation in most portfolios: one payroll number spread across six editions by a rule nobody wrote down.

The payroll has to land somewhere, and it is a big number

A permanent show team is the one large cost an organiser carries all year while the revenue arrives on six specific weeks. Contractors invoice against an edition. Freelance registration staff are booked for four days and coded to the show they worked. The twelve people who plan all of it get paid in January, when nothing is open and nothing is earning.

Scale gives a sense of the stakes even in businesses much larger than a single portfolio. RELX PLC, whose Exhibitions segment sits alongside its information businesses, says in its 2025 annual report on Form 20-F that "Our most significant cost item is staff costs", and puts those staff costs at 3,175 million pounds for 2025 against 3,120 million for 2024. Group revenue for 2025 was 9,590 million pounds, so staff cost runs at 33.1 per cent of it. RELX is largely an information business and that ratio is no benchmark for an exhibition organiser, though the ranking holds anywhere: the people line is the largest thing you will ever allocate, and it is the one most often allocated by habit.

What the standards decide, and what they leave to you

Two questions hide inside one. When does the payroll hit the profit and loss account, and which show's page carries it? The first has an answer in the standards. The second does not.

IAS 19, the IFRS Foundation's employee benefits standard, settles the timing. Paragraph 11 requires that "When an employee has rendered service to an entity during an accounting period, the entity shall recognise the undiscounted amount of short-term employee benefits expected to be paid in exchange for that service", and recognise it "as an expense, unless another IFRS requires or permits the inclusion of the benefits in the cost of an asset". January payroll is a January cost. You cannot park a show team's salaries on the balance sheet until the doors open on the grounds that the revenue turns up then.

That last clause has an exception in it, and IFRS 15 is the standard it points at. Paragraph 95 permits an asset from costs incurred to fulfil a contract where three criteria are all met: "the costs relate directly to a contract or to an anticipated contract that the entity can specifically identify", "the costs generate or enhance resources of the entity that will be used in satisfying (or in continuing to satisfy) performance obligations in the future", and "the costs are expected to be recovered". Paragraph 97 then lists what relates directly to a contract, and the first item is "direct labour (for example, salaries and wages of employees who provide the promised services directly to the customer)". Paragraph 98 sends the rest straight to expense, including general and administrative costs "unless those costs are explicitly chargeable to the customer under the contract".

Most organisers will fail the second of those criteria on most of their show team's time, and the amounts are small enough that nobody argues. The categories are still the right ones for the management pack, which is the useful part. An operations manager drawing the hall plan for a signed edition is doing work that traces to identifiable contracts. The same person spending February running a tender for a new registration supplier across all six shows is doing general and administrative work. One team, two categories, and the split turns on what the person did that week.

Four bases on one 650,000 payroll

Take six B2B shows run by one operations team of twelve, at a fully loaded annual cost of 650,000 including employer contributions. That averages 54,167 a head. Portfolio revenue is 10.5 million: March at 2.3 million, April at 1.6 million, June at 2.4 million, September at 2.1 million, October at 1.3 million and November at 0.8 million.

The two spring shows are 3.9 million of that, which is 37.1 per cent of the portfolio. They are also six weeks apart, and the April show moved venue this year.

An even split per edition gives each show 650,000 divided by six, which is 108,333, so the spring pair carries 216,667.

A revenue share gives the pair 3.9 of 10.5 million, or 37.1 per cent, which is 241,429. Call it 241,000.

A timesheet basis, with the team asked at year end where the time went, comes back with 63 per cent on the spring pair. That is 409,500. Call it 410,000.

Put those against the pair's contribution before the payroll charge, which is 1.28 million on 3.9 million of revenue. The even split leaves 1,063,333, or 27.3 per cent. The revenue share leaves 1,038,571, or 26.6 per cent. The timesheet leaves 870,500, or 22.3 per cent.

Five margin points between the highest and the lowest, from one payroll and three defensible rules. Where a show director's bonus is set on margin, that person has a direct financial interest in the basis, and they will find out what it is.

What does a timesheet percentage actually measure?

The timesheet feels like the honest basis. It is built on what people actually did, and it is the one I would trust least without a correction applied to it first.

Kaplan and Anderson set out the reason in Harvard Business School working paper 04-045, dated November 2003 and published the following year in Harvard Business Review as time-driven activity-based costing. Writing about the surveys that traditional activity-based costing depends on, they observed: "When people estimate how much time they spend on a list of activities handed to them, invariably they report percentages that add up to 100%. Few individuals report that a significant percentage of their time is idle or unused."

Hand a show team a list of six editions and ask them to divide up their year. The six numbers will total 100. Holiday will not appear. Neither will the four weeks on the registration tender, the health and safety review, the two people who spent a month on a system migration, or the ordinary slack any team has between two builds.

All of it gets pushed into the six buckets in rough proportion to what the person was thinking about, and the edition that felt busiest absorbs the most. The spring pair came back at 63 per cent because spring felt like 63 per cent of the year, and the March and April builds are the part anybody remembers in December.

Charging at a day rate built on practical capacity

The correction is the one Kaplan and Anderson propose. Estimate the capacity you are supplying, charge work out at a rate per unit of that capacity, and leave whatever nobody consumed sitting visible as unused capacity.

Twelve people at 225 available working days each, after holiday and public holidays, is 2,700 person-days a year. Kaplan and Anderson note that "Often practical capacity is estimated as a percentage, say 80% or 85%, of theoretical capacity", using a 40 hour week counted as 32. At 80 per cent, this team's practical capacity is 2,160 person-days. The rate is 650,000 divided by 2,160, which is 300.93 a day.

Now log days against editions for a year. That costs each person about thirty seconds a day and is the only new data collection in the exercise. Suppose the six shows absorb 1,780 person-days, of which 1,010 fall on the spring pair.

The spring pair is charged 1,010 times 300.93, which is 303,939. Call it 304,000. The other four take 770 days, or 231,716. The remaining 380 days of practical capacity, worth 114,353, went on portfolio work and slack and belongs to no edition at all. The three figures add back to the 650,000, give or take the rounding on the day rate.

Two things fall out. The spring pair's real share of logged show days is 1,010 of 1,780, which is 56.7 per cent against the 63 per cent the team reported. And 114,000 of payroll has stopped pretending to be a show cost. On the margin table above the pair now shows 976,061, or 25.0 per cent, sitting between the revenue share and the timesheet.

That 114,000 is the most useful figure the method produces, because it is the one somebody can act on. Either it is portfolio work that belongs in the pool covered by allocating central overhead to shows, or it is capacity you are carrying and could put somewhere else.

Which basis would I use?

The day rate on practical capacity, run once a year, with a revenue share as the fallback where nobody will log days.

What earns it the place is the residual. Every other basis forces the whole payroll down onto editions, which quietly asserts that a permanent team spends its entire year on shows. That assertion is false in every operations team I have seen, and here the size of it is 114,000, or 17.6 per cent of the payroll.

The even split has one real virtue worth naming: it cannot be gamed and it takes ten seconds. On six shows of similar size run by a team that genuinely rotates, it is the right answer and anything more elaborate is theatre. It stops working the moment show sizes diverge or a launch joins the calendar, because a launch eats far more team time than any of its measures suggest.

The revenue share is the one I would argue against hardest, and it is the most common of the three. It charges a show for being large, while operations time tracks complexity much more closely than it tracks revenue. An 0.8 million November show in a difficult venue with 140 small exhibitors can consume more operations days than a 2.4 million June show with 60 large stands in a hall the team has used for nine years.

Where this stops

A day log is self-reported too, and the optimism that inflates a year-end percentage will inflate a daily one in smaller increments. It is better because somebody recording today is remembering hours instead of months, and because the total is bounded by the calendar instead of by 100 per cent of a feeling. It remains an estimate. Kaplan and Anderson are relaxed about that, writing that the objective "is to be approximately right, say within 5-10% of the actual number, not to measure the capacity to four significant digits", and the same tolerance applies here.

The second limit is that none of this moves group profit by a penny, which is worth saying before anyone spends a quarter on it. Shifting 193,000 between the spring pair and the autumn four leaves the portfolio exactly where it was. The reason to care is that show margins get used for decisions, and a show closed on a margin distorted by a bad allocation takes its revenue out of the building while most of its allocated payroll stays behind.

The third limit is the one the standards flagged. Whether a show team's cost belongs in direct cost at all depends on whether you would really reduce the team over one soft edition, which is the question behind fixed versus variable event costs, and the answer for most organisers is no. A cost you would never cut is closer to overhead than to direct cost, whatever line of the event profit and loss structure it currently sits on.

This week, take the last edition you closed and ask the operations team a narrower question than the annual one: how many working days did this show take you, from the first planning meeting to the debrief? Six answers and a calculator give you a first day count, and the gap between that count and the percentage in your current allocation is the size of the problem, measured on your own event finance numbers.