Computing a trade show launch success rate from your own history
A trade show launch success rate is the count of your own launches that reached a stated survival test, divided by a census of every first edition you ran in the period. Both halves have to be published. Changing the survival definition moves the same portfolio's rate from roughly a third to nearly four fifths.
Somebody in the strategy session says the group is good at launches, and somebody else says most trade show launches fail. Both are stated with confidence and neither is attached to a number, so the meeting moves on to the next slide and the launch budget gets set on whichever of the two views the person holding the budget happened to agree with.
Your trade show launch success rate is computable from records you already have, and computing it takes about a day. The reason nobody has done it is that the answer depends entirely on two choices most groups have never made explicitly: what goes in the denominator, and what counts as survival.
Get those two right and the rate becomes usable. Skip them and the number will be too high, every time, in a predictable direction.
Build the census before you build the rate
Start with a list of every first edition your group ran under a new brand in the last ten years. Not every launch anyone remembers. Every first edition.
The gap between those two lists is the whole problem. In the portfolio I am using here, the group's own recollection produced twelve launches. The census, built by going through venue contracts and looking for a first invoice under a new event code, produced nineteen.
The seven missing ones fell into recognisable categories. Four were folded into an existing show after their second edition and are now described internally as feature areas or co-located zones, so nobody thinks of them as launches any more. Two were rebranded after a weak first edition and the group's records treat the second name as the launch, which quietly deletes the first attempt. One ran a single edition and was cancelled during planning for the second, and its file was archived under the venue rather than under the brand.
None of that is dishonesty. It is the ordinary way an organisation forgets, and it biases the rate upward because every deletion is a failure. A group whose remembered denominator is 12 with 7 survivors reports 58 per cent. On the real denominator of 19, the same 7 survivors give 37 per cent.
Two edge cases need a decision before you start counting, and the decision matters more than which way it goes. An overseas edition of a brand you already own behaves differently from a launch into a new sector, because it inherits an exhibitor list and a proven format, and mixing the two produces a blended rate that describes neither. Keep them in separate censuses and compute two rates, and hand the second one to the people working on an overseas edition of an existing show. The other case is a show you acquired and then relaunched under a new name, which is a rebrand rather than a launch and belongs in neither census. Write both rules down at the top of the sheet, because the temptation to reclassify a specific event once you can see the effect on the rate is genuinely strong.
What counts as survival?
This is the choice that moves the number most, and it has no correct answer, only a stated one.
Run the same nineteen launches against four definitions and watch what happens.
- Reached a second edition. 15 of 19 launches cleared this, which is 78.9 per cent. It measures whether the first edition was bad enough to stop, and almost nothing else.
- Reached a fourth edition at all. Three launches are still too young to judge, so the resolved base is 16. Eight of those 16 reached a fourth edition as an independent event, which is 50.0 per cent. The four that folded into an existing show after their second edition fail this test, which is the right answer and the one their champions dispute.
- Reached a fourth edition with positive contribution. Seven of 16, which is 43.8 per cent.
- Reached a fourth edition with contribution above the fully loaded cost of the team time it consumes. Five of 16, which is 31.3 per cent.
Same nineteen launches. Rates from 78.9 down to 31.3 per cent, a spread of 47.6 points, produced entirely by the definition.
My preference is the third one, contribution positive at edition four, because it is the point at which a launch stops consuming capital and starts being a show. The fourth definition is more demanding and probably more honest about opportunity cost, and if you use it you have to use it consistently when the rate is applied to a decision, which usually means a lower launch budget.
The three pending launches deserve their own treatment rather than a quiet exclusion. Published properly, this portfolio's rate is 43.8 per cent on 16 resolved launches, with three pending, and the ten year figure lands between 36.8 per cent if all three fail and 52.6 per cent if all three succeed. That range is the honest statement, and it is more useful than a single number because it shows how much of the answer is still unknown.
Why is your remembered number always too high?
Because the failures leave the organisation and the survivors stay in it.
The show that folded into a larger event has a champion who describes it as a successful integration. The rebranded one has a marketing story about repositioning. The cancelled one has nobody, because the person who ran it left two years later and their files went with them. Meanwhile the seven survivors are still running, each with a director who can tell you what edition it is on.
Measurement inconsistency of exactly this kind is a documented problem in the academic work on firm exit. Cefis, Bettinelli, Coad and Marsili reviewed 142 journal articles published between 1991 and 2020 in Small Business Economics in 2022, and described a field that has moved from treating exit as a homogeneous event signalling failure toward treating it as a heterogeneous one, with routes including voluntary liquidation, acquisition, flotation and bankruptcy sitting under the same word. If a body of 142 peer reviewed studies cannot agree on what exit means, an event group with no written definition will certainly not have converged on one by accident.
The exhibition industry already accepts this logic elsewhere. UFI requires that a UFI Approved Event has its visitor numbers, its net exhibition space and its exhibitor numbers independently audited under UFI's Auditing Rules, with the calculation standards and definitions set out in Annex 2 of the June 2021 version. Nobody in the industry thinks a self reported attendance figure is comparable across organisers without that. A self reported launch survival rate deserves the same scepticism, including from the person who computed it.
What the rate is precise enough to do
Three uses, and they get progressively more demanding of the number.
The first is ranking. If your rate is 43.8 per cent and the sector norm you assumed was 70, the launch budget is too large and the pipeline in the plan is too optimistic. That conclusion survives a wide margin of error.
The second is threshold testing. The build against buy decision resolves into a break even survival probability, and comparing your measured rate against that threshold is a legitimate use as long as the two are not close. If the threshold is 23 per cent and you measure 43.8, launching wins comfortably. If the threshold is 41 per cent, the comparison is inside the noise and something else has to decide.
The third is direct valuation, feeding the rate into a probability weighted pipeline. That is the most demanding use and the one where the small sample bites hardest, which is why the pipeline valuation should carry a chained stage by stage probability rather than a single overall rate. Stage probabilities are estimated from larger counts, because every launch contributes to the first stage even if only a few reach the last.
Where this stops
Sixteen resolved launches is a small sample and the rate carries an uncertainty most people will ignore once it has a decimal point on it.
Seven of 16 is 43.8 per cent. The standard error is the square root of 0.438 times 0.562 divided by 16, which is the square root of 0.0154, or 12.4 percentage points. At two standard errors the interval runs from about 19 to about 69 per cent. That is a very wide band, and it means the difference between a group reporting 44 per cent and one reporting 55 per cent is probably nothing.
Two consequences follow. Report the count rather than the percentage in internal documents, because 7 of 16 is harder to over read than 43.8 per cent. And pool where you honestly can, using stage probabilities across all 19 launches rather than a single end to end rate across the 16 resolved ones, since the first stage has 19 observations and the last has 16.
The rate also tells you nothing about why launches failed, and the why is where the improvement lives. A group failing at the first edition has a market selection problem. A group failing between editions two and four has a retention problem that looks much more like the evidence behind closing an existing show than like anything specific to launching.
This week, build the census. One row per first edition run in the last decade, taken from venue contracts rather than from memory, with the brand, the year, the edition count reached and whether contribution at the fourth edition was positive. Nineteen rows took an afternoon in the case above, and the gap between that list and what the group believed about itself was the most useful part of the whole exercise for the portfolio and synergy plan that followed.
Questions people ask about trade show launch success rate
- What is a realistic trade show launch success rate?
- There is no reliable published industry figure, and any number quoted without a definition and a denominator should be ignored. Compute your own. On a ten year census with survival defined as reaching a fourth edition with positive contribution, a rate somewhere between a third and a half is a common result for a mid sized portfolio.
- How do you count launches that were folded into an existing show?
- Count them in the denominator and treat them as non survivors under a survival test based on independent editions. A launch absorbed into a larger event after two editions did not become a standalone show, whatever the internal narrative says. Record the absorption separately so the group can see how often that outcome happens.
- How many launches do you need before the rate means anything?
- More than most groups have. Seven survivors out of sixteen resolved launches gives 43.8 per cent with a standard error of about 12 percentage points, so the honest range at two standard errors runs from roughly 19 to 69 per cent. Use the rate to rank decisions rather than to price them precisely.
Related reading
- Launch versus acquire a show and how to price the difference
- Valuing a show launch pipeline without paying twice for the same growth
- When to kill a trade show and the evidence that should decide it