Designing registration price increase tiers that pull the curve forward

The rate card came back from finance with four registration price increase tiers on it. Super early at 545, early at 595, advance at 645, standard at 695. Somebody had clearly enjoyed building it. Four dates, four prices, a clean fifty units between each step, and a note asking marketing to confirm the copy by Friday.

Nobody in the chain had asked the only question that decides whether the ladder does anything. Is fifty units, on a badge that costs five hundred and forty five, enough to make a person who was going to register in March register in January instead? On most B2B shows the answer is no, and the four tiers will produce four sets of deadline emails, four price changes in the registration platform, and a registration curve shaped exactly the way it would have been shaped anyway.

What is a price step supposed to buy?

A price ladder has one job that a flat price cannot do, which is to make waiting expensive. Everything else it appears to do is a side effect.

You are selling a badge whose value to the buyer does not change with the date. The show is the same show in January and in March. So the only reason to register early is that the price will be higher later, and the strength of that reason is the size of the step divided by the price the buyer is looking at. A step that is small relative to the base is a rounding error on a trip that also involves flights and three nights of hotel.

This matters commercially because early registrations are worth more than their face value to an organiser. They give you a readable file earlier, they let the sales team quote an audience number to exhibitors who are still deciding, and they reduce the size of the tail that operations has to guess at. A ladder that moves nothing has cost you real revenue on every registration that took the low tier while buying none of that.

Two published ladders, read as design

Two large shows publish their full price ladders, and both repay being read as design documents.

The NACS Show 2026 rate card runs three windows for a buyer on a full conference registration: 355 US dollars at the member early bird rate, 505 in the pre-show window running from 20 June to 25 September, and 655 on site after 26 September. The step is 150 dollars each time, held constant in absolute money. As a proportion of the price the buyer is facing, the first step is 42.3 per cent, since 150 over 355, and the second is 29.7 per cent, since 150 over 505. What the gap to the non-member rate on that same card is doing is a separate question and A20's.

The CONEXPO-CON/AGG 2026 registration policy publishes a four-step ladder for customer and dealer badges bought through an exhibitor's shared discount link: 153 US dollars from 5 August to 12 September 2025, 182 to 5 December, 211 to 2 March 2026, and 231 for show week. The steps are 29, 29 and 20 dollars, which is 19.0 per cent, then 15.9 per cent, then 9.5 per cent of the price in front of the buyer at each point.

Both ladders weaken as the show approaches. That is the pattern I would argue with. The people still undecided in the final window are the hardest to move and the most expensive to service, and they are being offered the smallest reason to stop waiting. On the CONEXPO ladder the buyer facing the last increase is looking at a 9.5 per cent rise, which is less than half the incentive the buyer faced in September.

There is a fair defence. CONEXPO's final window is five days long, so very few people are ever inside it, and the cost of getting that step wrong is small. The same defence does not cover the middle of a ladder, which is where most of the file sits.

The step has to be bigger than the cost of waiting

Before you set the steps, write down what waiting actually costs your registrant, because that is the number your step has to beat.

For a delegate whose employer pays, waiting costs almost nothing. It may even help, since the budget approval they are waiting on is more likely to exist in February than in October, and a registration made too early is a registration they have to defend if plans change. For a self-funding attendee, waiting costs the difference between the tiers and nothing else, because the ticket is not going to sell out.

So the step is competing against a genuine option value. My working rule is that a step below about fifteen per cent of the price in front of the buyer will not change a corporate decision. It will move some self-funders and it will be gratefully accepted by everybody who had already decided. The fifty unit step on a 545 base in the rate card above is 9.2 per cent, and it is on the wrong side of that line.

Yield across three tiers on 9,000 paid delegates

Take a paid conference of 9,000 delegates and put three tiers on it, using the NACS step size as a starting point: 595 early, 745 advance, 895 standard.

Suppose the file lands 4,200 in the early window, 3,100 in the advance window and 1,700 at standard.

Early: 4,200 times 595 is 2,499,000. Advance: 3,100 times 745 is 2,309,500. Standard: 1,700 times 895 is 1,521,500. Total registration revenue is 6,330,000, and the average yield per delegate is 6,330,000 divided by 9,000, which is 703.33.

Now price the two things the ladder is being compared against. A flat price at the middle tier of 745 across all 9,000 would produce 6,705,000, which is 375,000 more than the ladder delivered. A flat price at 895 would produce 8,055,000, which nobody would pay in full but which sets the ceiling.

So this ladder costs 375,000 against a flat mid price. That is the number the ladder has to earn back, and it earns it back in two ways only: registrations that would not have happened at 745 at all, and registrations pulled early enough to change what you could sell on the back of them. At 595, you need 630 additional delegates to cover the 375,000, since 375,000 divided by 595 is 630.3. On a 9,000 file that is seven per cent of the audience arriving purely because the early price existed.

Seven per cent is a testable claim. It is also, in my experience of watching these files, a generous one.

When does the fourth tier stop paying for itself?

Now add the super early tier the finance rate card wanted. Put it at 545 for the first eight weeks and assume it captures 1,400 of the 4,200 who would otherwise have paid 595.

The immediate revenue change is 1,400 times the 50 unit difference, which is a loss of 70,000. To break even, the tier has to produce 70,000 divided by 545, which is 129 delegates who would not have registered at any of the other three prices. Not 129 who registered earlier. 129 who would not have come.

The break-even count is small enough that the tier might clear it. The reason I would still refuse is that the 129 is invisible and the 70,000 is not. You will never be able to show that the fourth tier produced 129 incremental delegates, so the tier will be judged by its volume, which will look excellent, and it will be renewed every year on evidence that does not exist.

Then there is the administrative cost, which is real and is always understated. A fourth price is a fourth row in the registration platform, a fourth line in the deferred revenue schedule, a fourth variant in every group booking sheet, a fourth set of dates in the exhibitor invitation kit, and one more date on which somebody will ask for an exception. CONEXPO-CON/AGG's 2026 policy has a rule that exists precisely because of that pressure: discounts must be applied at the time of registration and cannot be applied once the registration fee has been paid. Every tier you add is another opportunity for someone to want that rule bent, and the most expensive version of bending it is moving a published date, which A15 covers.

The ladder I would build

Three tiers, with steps of at least twenty per cent of the price the buyer is looking at, and the largest step last.

Three windows are enough to create urgency twice, which is as many deadline campaigns as a marketing team can run well. Twenty per cent is above the threshold where a corporate approver notices. Putting the largest step last inverts what both published ladders above do, and it is the change I would most want to test, because the late window is where the pull-forward opportunity actually lives. Whether a given cutoff moved anything is measurable on the daily curve, which is A12's method, and it is worth having that reading before you touch the steps.

PCMA reported in April 2024, drawing on Maritz research, that time-based discounts are open to a specific objection: you might be giving away money to the very people who would attend your show regardless. Sizing that on your own file is a separate measurement and A14's. The same article set out the alternative, which is to tier by what the registration contains rather than by when it was bought. One price for exhibit hall access, a higher one for the full conference, and an intermediate tier defined by session count. That ladder discriminates on how much of the event a person wants, which is a real difference between two buyers, and it does not hand a discount to a repeat attendee for behaviour they were going to exhibit anyway.

I would run both axes. Date tiers to shape the curve, content tiers to capture the delegates who value the conference programme, and no more than three steps on either.

Where this stops

The yield arithmetic above assumes the distribution across tiers stays put when you change the prices, and it will not. If you widen the steps, some share of the early window buyers move up a tier rather than pay more, and some share of the standard tier simply does not register. Neither effect is knowable from your own file, because you have never run the other ladder.

That is a genuine limit and it is why I have given a rule of thumb rather than an optimisation. Getting past it needs a real price test, which means two randomly assigned groups seeing different ladders in the same edition, and most organisers will not run one because showing two prices for the same badge is difficult to defend if a registrant notices. The honest fallback is a between-editions comparison with everything else held as still as you can manage, which takes two years and gives you one observation.

The other limit is that the ladder only governs the population that pays. On a show where the expo floor is free and the conference is paid, the tiers touch a minority of the file, and the curve you are trying to pull forward is mostly made of registrations that no price is acting on.

Take last edition's paid registration file and add one column to your attendee analytics extract: which tier each registration was bought at, and the date it was created. Compute the average yield per delegate, then compute what a flat price at your middle tier would have produced on the same volume. The difference between those two numbers is what your ladder cost, and it is the number the ladder has to justify.