Self service task completion rates by task type across the onboarding window

Two weeks out, the operations lead says onboarding is at 74 per cent and everyone nods, because 74 sounds like a show that is mostly ready. Nine days later the compliance list has 61 exhibitors with no valid certificate of insurance and 88 with no W-9, and nobody can explain how a 74 per cent portal produced that.

Self service task completion behaves differently task by task, and a single blended rate averages the behaviours away. Company profile copy lands months out. Insurance lands in the final fortnight. Tax forms land after that, or after somebody phones. Those are three separate processes wearing one number.

ExpoFP (2026), laying out an 18 month planning timeline, has the exhibitor kit going out four months before the show. That gives a working window of roughly 120 days, and the useful question is what happens across those 120 days for each task separately.

One completion rate for the whole portal hides three problems

The blended rate is a weighted average of curves with different shapes, and averages of differently shaped curves are hard to reason about.

Consider what the show can do about each. A task that completes early and completes fully is finished, and no reminder should be sent about it. A task that completes late but reliably needs the deadline moved or the exhibitor warned earlier. A task that never completes fully needs somebody to decide whether it is really required. A task with almost no completions at all may simply have opened in the portal last week, which is a configuration fact rather than an exhibitor behaviour.

The blended rate treats all four the same way, which is why it produces the same operational response every year: send everyone a reminder about everything. That response is expensive, it trains exhibitors to ignore your email, and it does not move the tasks that matter.

Splitting by task takes one group-by clause and changes what you can say in the room.

Plot each task against days before move in

Build the curve as a cumulative percentage. For each required task and each day in the window, count the exhibitors who have completed and had that task accepted, divide by the exhibitors the show requires it from, and plot against days before move in with the axis running down to zero.

Two choices decide whether the curves are comparable across editions.

Index to days before move in, never to calendar dates. Show dates shift, and the whole point is comparing this edition's insurance curve against last edition's at the same distance from the hall opening.

Fix the denominator at the exhibitors who owe the task, recomputed weekly as booths sell. The completed-task denominator problem is the same one here, and getting it wrong will make every curve drift downwards through the season as the book grows.

Record when the task list itself changes. A rule added in week nine creates a task that could not have been completed in week one, and a curve drawn over a moving definition is unreadable. Keeping every rule under version control with an effective date is what lets you mark the curve at the point the requirement appeared.

What do the curves actually look like?

Here are three, with figures invented so the shape and the arithmetic can be followed.

Company profile, as a percentage of the 400 exhibitors who owe it: 44 per cent complete at 90 days before move in, 71 at 60 days, 88 at 30 days, 95 at 7 days, and 96 at the end.

Certificate of insurance: 9 per cent at 90 days, 24 at 60, 58 at 30, 84 at 7, 91 at the end.

W-9: 4 per cent at 90 days, 11 at 60, 33 at 30, 64 at 7, and 78 at the end.

The company profile curve is steep on the left and flat on the right. The W-9 curve is flat on the left and steep on the right. The insurance curve sits between them and moves hardest between 60 and 30 days.

A second reading makes the difference concrete. Take the share of eventual completions achieved in the first 30 days of the window. Company profile got 44 of its final 96, which is 45.8 per cent of everything it will ever get, done in the first quarter of the window. Insurance got 9 of 91, which is 9.9 per cent. The W-9 got 4 of 78, which is 5.1 per cent. On the same portal, with the same 400 companies, one task is nearly half finished while another has barely started.

The number to read off a completion curve

Curves are hard to put in a report and easy to argue about, so compress each one to a single figure: the day at which half the eventual completions have happened.

Company profile crosses 50 per cent between the 90 day reading of 44 and the 60 day reading of 71. It needs 6 more points out of the 27 gained across that interval, so 6 over 27 of 30 days is 6.7 days, putting the crossing at about 83 days before move in.

Certificate of insurance crosses between the 60 day reading of 24 and the 30 day reading of 58. It needs 26 of the 34 points gained, so 26 over 34 of 30 days is 22.9 days, and the crossing sits at about 37 days out.

The W-9 crosses between 30 days at 33 and 7 days at 64. It needs 17 of 31 points across 23 days, which is 12.6 days, putting it at about 17 days before move in.

The half completion days are 83, 37 and 17, a 66 day spread across three tasks on one portal, and this is the single most useful thing the curve gives you, because it turns into a schedule. Any reminder about the company profile sent later than about 90 days out is arriving after most of the work is done. Any reminder about the W-9 sent before 40 days out is arriving before anyone has thought about it.

Track those three numbers edition over edition. If the insurance half completion day moves from 37 to 29, the show is drifting later and the compliance crunch at move in will be worse than last year, and you know it four months early.

Where should the reminder go?

Put it just before the steep part of each curve, which is a different date for each task.

The mechanism is worth being precise about. A reminder mostly pulls forward work that was going to happen anyway, and it converts a small share of never-going-to-happen into happened. Both effects are largest when the exhibitor is close enough to the deadline to care and far enough from it to act.

On the numbers above that implies a profile reminder around 100 days out, an insurance reminder around 50, and a W-9 reminder around 25, with a second W-9 pass at 10. Sending all three to everyone at 60 days, which is what a blended rate encourages, wastes the profile reminder entirely and fires the W-9 one at a population who will file it away.

You can test this directly. Hold out a random 15 per cent of exhibitors from one task's reminder for one edition, compare the two half completion days, and you have measured the reminder rather than assumed it. The cadence design that follows from a result like that belongs to its own post, and the tagged inbound email volume will tell you whether the reminder generated questions as well as completions.

Tasks with a third party in them behave differently

The flat left half is a signature, and it usually means the exhibitor cannot act alone.

A certificate of liability insurance is issued by the exhibitor's broker, so the exhibitor's part is a request and a wait. The IRS Form W-9, currently the March 2024 revision of the Request for Taxpayer Identification Number and Certification, needs a signature from somebody in finance who has no interest in your show. A booth layout drawing needs the stand builder. None of those complete quickly no matter how good the upload screen is.

For those tasks, move the request earlier and reminders matter less. Ask for the certificate at contract signature, when the exhibitor is already talking to somebody about money, and the broker request goes out four months earlier. Prefill the W-9 fields the show already knows, so the only act left for the exhibitor is a signature. Publish the layout drawing requirement in the booth confirmation as well as in the portal, because the stand builder never sees the portal.

Where a task genuinely has no third party and still shows a flat left half, the screen is the suspect, and ranking screens by email per exhibitor will confirm it or clear it in a month.

Where this stops

These curves describe a population and say nothing about an individual. A first time exhibitor's certificate curve and a fifteen year veteran's are different shapes, and blending them produces a half completion day that describes nobody. Split by tenure before you draw conclusions about drift, because a show that has just sold 80 new stands will see every curve shift right for reasons that are about the mix rather than the process.

The curves also cannot separate willingness from opportunity. A task that opens in the portal 60 days out cannot show completions at 90 days, and its half completion day will look late in a way that reflects your own release schedule. Record the date each task opened alongside the curve, or you will diagnose exhibitor behaviour when you are actually looking at your own release schedule.

The last limit is the end point. Half completion day depends on knowing the eventual total, which you only know after the show. During the season you are estimating it, and early in the window that estimate is poor. Use the previous edition's final rate as the denominator for in-season readings, mark it as an estimate, and recompute properly once the show closes.

Pull last edition's task completion timestamps, index them to days before move in, and compute the half completion day for your three heaviest tasks. If they are more than a month apart, your reminder schedule is currently wrong for at least two of them, and the fix is a calendar change rather than an onboarding rebuild.