The post show rebooking window and how fast intent decays

The post-show rebooking plan lands in your inbox on the Tuesday after teardown. Twelve weeks of outreach, three touches per account, evenly spread, with the heavy push scheduled for week six because that is when the team has recovered from the show and the floorplan for next year is ready.

Week six is too late by a wide margin. The window in which exhibitors actually sign is short and it is front loaded, and the size of the planning error is measurable on data you already hold, because every contract in your system carries a signature date and every show has a closing date.

Count renewals by week, on a base that shrinks

The measurement that answers this is a weekly hazard: of the accounts that had not renewed at the start of week t, what share renewed during week t. The denominator changes every week, which is the whole point and the thing most reporting gets wrong.

Take our industrial show again. 640 exhibiting companies, 402 of them signed during show week, which is an onsite rebooking rate of 62.8 per cent, so 238 accounts leave the hall unrenewed. Those 402 were worked through a funnel of their own, and the post show rebooking window starts where that funnel ends. The 238 are the base, and here is what the eight weeks after close produced.

Week one, 43 signed from a base of 238, which is 18.1 per cent. Week two, 31 from 195, or 15.9 per cent. Week three, 22 from 164, 13.4 per cent. Week four, 16 from 142, 11.3 per cent. Week five, 11 from 126, 8.7 per cent. Week six, 8 from 115, 7.0 per cent. Week seven, 6 from 107, 5.6 per cent. Week eight, 3 from 101, 3.0 per cent.

Eight weeks produced 140 contracts from 238 accounts, 58.8 per cent of the base. The first fortnight produced 74 of those 140, so 52.9 per cent of everything the whole window delivered arrived before most teams have finished writing the post-show report.

Run the same figures against an evenly spread budget and the mismatch is stark. Eight weeks of equal spend puts 25 per cent of the money against 53 per cent of the outcome, and the last four weeks take half the budget to deliver 28 contracts.

Why does the rate fall when nobody has changed their mind?

The tempting reading is that exhibitors cool off. Enthusiasm from the show floor fades, the lead scans disappoint, a budget round intervenes. All of that happens, and none of it is needed to produce the curve above.

A base made of accounts with different signing propensities will show a falling weekly rate even if every individual account's propensity is fixed forever. The keen ones convert early and leave the base. What remains is progressively colder, so the average of the remainder falls, and the aggregate rate falls with it.

Work it with two segments. Of the 238 unrenewed accounts, say 90 are keen, with a 35 per cent chance of signing in any given week, and 148 are cold, at 5 per cent a week. Nobody's number ever changes.

Week one: expected signings are 90 times 0.35, or 31.5, plus 148 times 0.05, or 7.4. That is 38.9 from 238, a rate of 16.3 per cent. Remaining, 58.5 keen and 140.6 cold.

Week two: 20.5 plus 7.0 is 27.5 from a base of 199.1, a rate of 13.8 per cent.

Week three: 13.3 plus 6.7 is 20.0 from 171.6, 11.7 per cent.

Week four: 8.7 plus 6.3 is 15.0 from 151.6, 9.9 per cent.

The rate has fallen from 16.3 to 9.9 in a month and the two segments are exactly as keen as they were on day one. What changed is the mix. Keen accounts were 37.8 per cent of the base in week one, 90 of 238. By the start of week four they are 24.7 of 151.6, which is 16.3 per cent.

This matters operationally because the two explanations imply opposite actions. If enthusiasm is decaying, speed is everything and a fast follow-up genuinely rescues sales. If the mix is sorting, speed buys you the keen accounts a few days earlier and does very little for the cold ones, and the money would be better spent on a different treatment for the residue.

Fitting the curve properly

The two-segment version is a teaching device. The general form replaces the two segments with a continuous distribution of individual signing probabilities, and the standard treatment is the shifted-beta-geometric model that Fader and Hardie set out in the Journal of Interactive Marketing in 2007.

Their setup has churn as the event and retention as survival, and the mechanism transfers directly with signing as the event and the unrenewed pool as the survivors. Each account has its own per-period probability, those probabilities follow a beta distribution across the population with two parameters, and the observed aggregate rate is what falls out. The paper makes a point I would repeat to any analyst reaching for a trend line instead: a curve fitted to past aggregate rates has no story attached to it and projects badly, while a two-parameter probability model with a plausible account of the process behind it forecasts the same data considerably better and can be estimated in a spreadsheet by maximum likelihood.

Two parameters is the appeal. You will have eight or ten weekly observations per edition, which is not enough data to support anything elaborate, and the beta mixture gives you a fitted curve, a projection past the end of your observed window, and an estimate of how heterogeneous your base is. That last quantity is the interesting one. A base that sorts sharply has a small keen segment worth chasing hard and a large residue worth a different approach entirely.

What this will not give you is a date for a named account. The hazard describes a population, and putting a time to event estimate on one exhibitor is a survival model at account level, built on a different set of inputs.

What should the curve change about the outreach plan?

Once you have the weekly hazard, planning stops being a matter of taste.

The first fortnight gets the people. Every rep who worked the show floor should be on the phone in week one, working accounts they personally met, while the conversation is still specific and both sides remember the stand number. Nothing you can do in week six substitutes for this.

Weeks three to five get the sequenced follow-up, the floorplan release and the pricing deadline, because at a 13.4 and 11.3 per cent weekly rate there is still real volume moving.

From about week six the economics change. At 7.0 per cent falling to 3.0, a rep working the residue at six calls a day is converting under half a contract a day, and the same rep is now competing with pre-show work for the next edition. This is where an automated path earns its place, and where I would stop counting the outreach as renewal work and start counting it as reacquisition.

There is a budget argument here that lands with finance, particularly in a year when UFI's 37th Global Exhibition Barometer, concluded in June 2026 across 466 companies in 59 countries and regions, found a majority at 54 per cent expecting stable operating profits. Moving spend from week seven to week one costs nothing and is not a headcount request, which is the sort of change renewal intelligence work can pay for on its own.

The confound you cannot ignore

Your own outreach shapes the curve you are measuring, which makes the hazard a description of a process you control rather than a natural law of exhibitor behaviour.

If your team makes 400 calls in week one and 60 in week six, of course week one converts better. The measured decay then contains a real sorting effect, a real intent effect, and a large effect from your own calling pattern, and the three are entangled.

The clean answer is to hold a randomly selected slice of the unrenewed base out of the week-one push and work them on the normal schedule from week three, then compare. A hundred accounts held back for two weeks costs you a few contracts and buys you the first honest estimate of what the early push is actually worth. Ascarza and colleagues make the broader version of this argument in Customer Needs and Solutions in 2018: predicting who is at risk is a different question from deciding who to target, and a retention programme that never separates the two ends up spending on accounts that would have converted with no help at all.

If a holdout is politically impossible in a renewal cycle, the cheap substitute is to record contact events with timestamps and put contacts per account per week next to the hazard on the same chart. Anyone reading it can then see whether the curve is following the base or following the calls.

Where this stops

The window is a property of an edition, and it moves.

A show that releases its floorplan two weeks after close has a different curve from one that releases on the closing day, because a large share of accounts cannot sign until they can see where they would be. If you change that release date, the hazard shifts and last year's curve stops applying. The same is true of a price rise announced at a fixed date, a venue change, or a competing show that opens its own rebooking six weeks after yours closes.

Eight to ten weekly observations from one edition is also a thin dataset, and the later weeks carry small counts where three contracts against 101 accounts is a rate with a wide interval on it. Pool two or three editions of the same show before you act on the tail, and keep shows separate, because a January show and a September show do not share a calendar.

The honest summary is that the shape holds up well and the exact numbers do not travel. Any base with mixed propensities produces a falling weekly rate, so you can expect the front-loading conclusion to survive on almost any show. Whether your own week six is worth funding is a question only your own contract dates answer.

Pull every contract signed for the next edition, subtract the show close date from each signature date, and bucket the result into weeks. That histogram takes twenty minutes to produce and it tells you immediately whether your current outreach calendar is pointed at the weeks where your exhibitors actually sign.