Measuring exhibitor onboarding cycle time from countersignature to show ready status

Somebody asks how long it takes to onboard an exhibitor. The operations lead says about three weeks. The coordinator who works the chase list says it takes all year. They are both describing the same 268 companies, and both of them are right, which is what a badly reported exhibitor onboarding cycle time does to a conversation.

Three weeks is the median. All year is the tail. A single average sits somewhere between them and describes nobody.

What the clock starts and stops on

Cycle time needs two dates per exhibitor, and both of them need arguing about once and then writing down.

The start is the countersignature. The exhibitor signed at some earlier point and the organiser signed at some later one, and only the second date represents an obligation on both sides. Using the exhibitor's signature date makes your own countersigning delay invisible, which is convenient and wrong.

The end is show ready, which is a definition you have to write. Mine is the first date on which the exhibitor met every published requirement at once: a certificate of insurance accepted, tax form on file and matched, any stand design approved, directory listing published. The words "at once" carry the weight. An exhibitor whose certificate expired in October and was replaced in December was not show ready in October, and a system recording the first arrival of each document rather than their simultaneous validity will date them months early.

One person should own that definition and it should live in the same document as the deadline calendar, because a show ready rule invented by an analyst during reporting week will differ from the one the operations team has been working to all year.

Everything between those two dates is elapsed calendar time, including weekends, holidays and the six weeks when the show was between editions and nobody was working on it. Business days are tempting and I would avoid them, because the exhibitor experiences calendar days and the deadline is a calendar date.

Why report P50 and P90 together?

Take the 268 exhibitors above, all of whom reached show ready. Sort their cycle times ascending. The median, P50, is 21 days. The 90th percentile, P90, is 96 days. The mean is 34 days, which is above the median and below the point where the trouble starts, so it manages to describe neither group.

Those two numbers together say something a single number cannot. Half the floor completes in three weeks, so the process, the paperwork and the portal all work. Then 27 exhibitors, the last decile, take between 96 days and whatever the maximum is, and those 27 companies are where the coordinator time goes in the final fortnight.

The operational reading is direct. A show with a P50 of 21 and a P90 of 96 has a chasing problem confined to the last decile, and redesigning the portal for everyone would be effort spent on the 241 exhibitors who are already fine. A show with a P50 of 74 and a P90 of 110 has a different disease entirely, because the middle of its floor is slow and the fix is structural.

Computing the percentiles by hand

Percentile definitions vary between tools and the differences matter at small counts, so pick one and write it into the definition alongside the metric.

The nearest rank method is the one to explain in a room. Sort the 268 cycle times ascending. For the 90th percentile, multiply 0.90 by 268, which is 241.2, then round up to 242. The 242nd value in the sorted list is P90. For the median, 0.50 times 268 is 134, and with an even count you take the mean of the 134th and 135th values.

Do that on a spreadsheet once and you will notice something the median hid. On our show the 242nd value was 96 days and the 268th, the maximum, was 214 days. The gap between P90 and the maximum is wider than the gap between P50 and P90. There is always a company that signed in February and published a profile the week of the show, and a percentile keeps it from dominating the headline while a mean does not.

Does a fixed external deadline change the measurement?

It does, and this is the part that trips up teams comparing cycle time across shows in a portfolio.

Onboarding has a moving start and a fixed end. Exhibitors sign across a window of many months and the deadlines land on the same date for everyone. The American Urological Association's published exhibitor requirements for its 2025 annual meeting set a single certificate deadline of 21 March 2025 and required coverage running from the first day of the exhibitor move-in period to the last day of the move-out period. Every exhibitor faced that date, whether they signed 300 days out or 40.

The consequence is arithmetic. An exhibitor who signs eleven months out has eleven months of available slack and will often use most of it, so their cycle time is long while their behaviour is fine. An exhibitor who signs six weeks out cannot post a 96 day cycle time even if they try. Cycle time measured from countersignature is therefore partly a measure of when your sales team closed, which is not what anyone thinks they are looking at.

So report a second figure next to it: days between show ready and the published deadline, signed so that positive means early. On our show the median exhibitor was show ready 34 days before the deadline and the tenth percentile of that distribution was 3 days after it. The two views disagree about who your problem exhibitors are, and the second one is the view that predicts what happens at move-in.

What sets the floor on the whole window

There is a limit to how much cycle time you can compress, and your own calendar sets it.

ExpoFP's trade show planning checklist, published in 2026, lists the floor plan lock as an event director milestone six months before the show and the exhibitor kit as an operations milestone at four months. Those two dates bound most of what onboarding can do. An exhibitor cannot order services from a kit that has not opened, and a stand position cannot be confirmed before the plan is locked, so an exhibitor who countersigns nine months out has three months of unavoidable waiting built into their cycle time before anybody is slow at anything.

Which means a raw cycle time distribution across a full selling window is measuring your milestone calendar as much as your operation. If you want the operational number, restart the clock at the later of countersignature and kit opening. On our show that single change moved the median from 21 days to 18 and moved P90 from 96 to 61, because most of the extreme values were early signers waiting for a kit, and few of them were late exhibitors ignoring reminders.

Both versions are worth having. The first is the exhibitor's experience of you. The second is your team's performance, and confusing them will send you chasing a tail made of your own release schedule.

The exhibitors who never finish

Every cycle time above was computed on exhibitors who reached show ready. The ones who never did have no cycle time at all, and dropping them silently is the most common way this metric flatters a show.

Those records are right censored: you know the exhibitor had not finished by the date you looked, and you do not know when or whether they would have. Excluding them biases the distribution downward, because the slowest cases are exactly the ones missing. Including them at their observed elapsed time biases it downward too, since their observation was cut short.

There are proper statistical treatments for censored durations and they are worth the effort on a large portfolio. On a single show with a few hundred exhibitors, the reporting discipline gets you most of the value for none of the machinery. The honest minimum is to publish the completion share alongside the percentiles. On our show, 268 of 291 contracted exhibitors reached show ready, so the percentiles describe 92.1 per cent of the cohort and 23 companies are outside the measurement entirely. Anyone comparing 21 days against another show needs to know whether that other show's median covers 92 per cent of its floor or 60 per cent. The share of exhibitors clearing each individual gate is the funnel stage view, and where those days accumulate inside the process is the state residency view.

Where this stops

Cycle time is silent about quality, and it can be improved by lowering your standards. A show that stops requiring an additional insured endorsement will see its median drop within one edition, and the drop will look like an operational win in every chart. Read cycle time next to a compliance rate you could defend in an audit or you have built an incentive to accept worse documents faster.

The second limit is the retrospective clock. If your show ready definition changes, every historical cycle time changes with it, and most teams discover this after publishing a year on year comparison. Recompute history under the new definition or publish a break in the series, and say which you did.

The third is that percentiles need a reasonable number of exhibitors. On a 40 exhibitor show, P90 is the 36th value and moves by a week if one company sends a certificate late, so report the raw sorted list and stop pretending the percentile is stable.

Take last edition's file this week, derive two dates per exhibitor, and sort the elapsed days ascending in a spreadsheet. Read off the middle value and the 90 per cent value, then count how many exhibitors have no end date at all. Those three numbers are a more useful description of your onboarding than any average anyone has quoted you.