Registration target setting that survives contact with the show director

The number usually arrives in July, eight months out, in a budget meeting that has already run long. Twelve thousand. The show director says it the way you say a number you have said before, and nobody asks where it came from, because everyone in the room half suspects the answer. Registration target setting, at most shows, is that moment and nothing else.

It came from last year's 10,800 plus a bit. Or from the revenue plan divided by an assumed yield per registrant. Or from a competitor's press release that the board also read. By November it is on a slide with no note attached, and by February the audience acquisition team is being asked why they are behind on a number that was never derived from anything.

A forecast and a target are different objects

A forecast is a claim about what will happen if nothing changes. A target is a claim about what the business intends to make happen and what it is prepared to spend trying. They answer different questions and they should never be the same number by accident.

When a show sets its target equal to its forecast, the target is doing no work. When a show sets its forecast equal to its target, which is much more common, the forecast has stopped being a measurement and become a restatement of ambition.

I would keep both in the same table, in two columns, and I would make it a rule that only the forecast gets revised in flight. The target moves once, in public, with a reason. Everything else is pacing commentary, which is A1's subject, and turning a mid-campaign read into a number for show open is A9's.

What should a registration target be built from?

Five editions of final registration counts, in order: 10,200, 10,800, 9,900, 11,100, 11,000. The median is 10,800 and the observed range runs from 9,900 to 11,100.

That band is the show's demonstrated capability under the campaign budgets, sales effort and market conditions of those five years. It includes a bad year and a good one. Building it properly, as a band with percentile edges at each day out, is A7's job. For target setting you need only the endpoints and the middle.

The band matters because it converts an argument about ambition into an argument about distance. A target of 11,200 sits inside a range the show has already reached twice. A target of 12,000 sits 8.1 per cent above the best edition on record, which is 12,000 divided by 11,100. Those are different conversations and they deserve different amounts of scrutiny.

Split the baseline before you add anything to it

A single registration total hides the only decomposition that matters for target setting. Split each edition into registrants who appeared at a prior edition of the same show and registrants who did not. That split is the first thing any attendee analytics pack should carry, because a total on its own cannot tell you which half of the file moved.

For the same five editions, returning registrations ran 6,400, 6,600, 6,100, 6,500 and 6,500. New registrations ran 3,800, 4,200, 3,800, 4,600 and 4,500.

Returning is stable. It sits between 6,100 and 6,600, and the median is 6,500. New is the volatile half, and its best edition ever produced 4,600.

Now put the target through that split. If you commit to 12,000 and returning holds at its median of 6,500, new registrations have to reach 5,500. The show has never produced more than 4,600 new registrants in a single edition. The target therefore requires 900 more new registrants than any prior edition has delivered, which is a 19.6 per cent improvement on the best year the acquisition team has ever had.

That sentence is the whole exercise. It is not an objection to the target. It is the target, stated in the only currency the people who have to hit it can act on.

The split also tells you which lever the target is really pulling. If returning had been drifting down, say 6,800 to 6,100 across the five editions, then a 12,000 target built on a returning median is quietly assuming you reverse a decline as well as break a new registrant record, and those are two separate programmes of work with two separate owners. Retention work moves slowly and lands a year later. New registrant acquisition responds inside a campaign cycle. A target that needs both to move in the same eight months is usually a target that needs one of them to move a lot and the other to be honest about not moving at all.

Price the increment before anyone agrees to it

The 900 has a cost, and the cost is higher than the average.

Suppose the last two editions delivered new registrations at a blended 42 units of acquisition spend each, across search, paid social, list rental and partner promotion. The 4,600 new registrants of the best edition therefore cost roughly 4,600 times 42, or 193,200 units.

The incremental 900 do not come from the same place. The audiences that convert at 42 have already been worked. The next 900 come from colder lists, wider geography and job functions the show has not previously sold to, so assume a cost per new registration somewhere between 60 and 75. Take 65. The increment costs 900 times 65, or 58,500 units.

So a target that is 11.1 per cent above the five edition median implies a 30.3 per cent increase in new registrant acquisition spend, since 58,500 divided by 193,200 is 0.303. If the budget line is not moving, the target is a wish.

Worth holding alongside this: CEIR's 2026 Marketing Spend Decision Report, based on a survey of 362 B2B exhibitors conducted in late 2025, found exhibitions taking 40.8 per cent of exhibitor marketing budgets, with 47 per cent of exhibitors expecting to do the same number of shows next year, 28 per cent expecting to add shows and 83 per cent planning to keep their booth size the same. The people buying your floor are steady. A target that assumes the audience side can grow 11 per cent while the exhibitor side holds flat is assuming you win share, and share has to be taken from a named competitor with a named tactic.

Write down what would have to be true

The output of target setting should be a short document that anybody can check later. Not a forecast, not a narrative. A list of conditions, each with an owner and a date by which it becomes knowable.

• Returning registrations hold at 6,500 or better. Owner: audience acquisition. Checkable at day minus 90 by comparing returning registrations at the same days out across the prior five editions.
• The two new list sources deliver at least 180,000 reachable, permissioned contacts by 1 October. Owner: marketing operations. Binary, and knowable early.
• Cost per new registration on cold audiences stays at or below 75 units. Owner: performance marketing. Knowable after the first 300 registrations from cold sources.
• The 58,500 unit increment is ring fenced and not reallocated to cover a sponsorship shortfall. Owner: finance.
• Registration operations and the venue can process and badge 12,000. Owner: operations. Sounds trivial until the year it is not.

Every one of those is falsifiable before the show opens. That is the point. A target with five written preconditions fails loudly in October, when you can still do something, instead of quietly in February, when you cannot.

Which error is worse, a target set too high or too low?

Both directions hurt, and they do not hurt equally.

A target set too low wastes money that was available and lets a show under-deploy a campaign that would have worked. It is a real cost and it is recoverable, because next year you set a higher one.

A target set too high does something more damaging, because the sales team sells the floor against it. Exhibitors buy space on an audience promise, and the gap between the promise and the turnstile is the number they carry into the rebooking conversation. CEIR's Performance Benchmark Playbook, second edition, published in 2026 and covering B2B exhibitions with 200,000 or more net square feet of paid exhibit space, reported a median gross revenue of 12.5 million US dollars and an average net profit margin of 55 per cent, with 80 per cent of organisers reporting profitability. A show with that shape can usually afford to fund a genuine acquisition increment. What it cannot afford is a floor sold twelve months ago against an audience number it missed by 1,200.

So my preference is asymmetric. Set the target at the top of what the written conditions support, fund it properly, and refuse to move it up in October because the first quarter looked good.

Where this stops

The method assumes the show you are targeting is the show your history describes. Change the qualification rules, move the venue, shift the dates by a season, absorb a competitor's event, or drop a free expo tier, and the baseline band describes something that no longer exists. In those years the honest answer is that you have a sample of zero and the target is a judgement, which you should say out loud instead of dressing it in percentiles.

The split between returning and new is also only as good as your identity matching. If you decide who is returning by exact email match, you will systematically classify job changers as new, which understates returning and overstates new. On the numbers above, if the true returning count is 6,900 rather than 6,500 because 400 people changed employers, then the best ever new registrant count is 4,200 rather than 4,600, and the gap to 12,000 is 1,300 instead of 900. The target got harder because your matching got better, which is an uncomfortable conversation to have mid campaign and a much easier one to have before the number is signed.

Take an afternoon this week and pull the final registration file for the last five editions. Tag every row as returning or new against the prior edition using the best identity match you have, and write the two numbers on one line per edition. If the target on the budget slide implies a new registrant count outside that range, you now have the specific number the argument is actually about.