Home TypeNews Phil Soar: are visitor conversions falling?

Phil Soar: are visitor conversions falling?

by EN

EN guest editor and chairman of CloserStill media, Phil Soar, looks at causation, correlation and co-incidence and asks if visitor conversion rates are falling?

Three times in the last month I have sat in meetings where worried marketing managers have fretted about their pre-reg conversion rates (for those in ignorance, this is the percentage of visitors who have registered for a show and who actually attend).
There has long been an industry belief that around 50% of those who register attend, though shows operating in a strong vertical (dentists, clothing) have been used to more like 65%. I have never seen any proof that 50% ever was the real average, but there is no way of knowing that.

Do we need to panic if we are looking at 35%?

It is hard to find reliable figures – very few of our companies will even whisper statistics which do not show them in a good light. But anecdotal evidence suggests that average conversion in the last 6 months has been around 35%.
For the next year we can assume that square metre and visitor numbers are likely to be below 2019, but it has always been the case that organisers are reluctant to admit to any number which is not an increase. Hyve and Informa have to release information because they are publicly quoted companies, and the RX figures tend to be lost in Elsevier’s generalised statements.
If you believe in causation, then surely it suggests people are more reluctant to attend? but are trade shows really losing their appeal?
If you believe in causation, then it appears that fewer people want to attend trade shows. I stress appears. If fewer of them turn up, then trade shows must be losing their appeal? And even when they have registered their interest, they still don’t come?

Isn’t that what the numbers are saying – we are losing our appeal?

But I very much doubt that this is the case, and here we get into the interesting debate about causation, correlation and coincidence.
Most of the shows which have run over the last six months – and the next six – have not appeared for two or even three years. But, by and large, they have retained the registrations which had already been received for events which were cancelled two years ago. In addition, they have probably kept websites open and have been collecting potential visitors for at least the last 18 months (largely because none of us were sure when we might be able to run the next show).
Add to that the fact that only 74% of British adults have been vaccinated. Now you can argue that the other 26% are not likely to be the core audience of an NEC trade show. Well, perhaps not. But there must be some overlap and we must assume that one quarter of the population is likely not yet confident about large gatherings.

The strange correlation between the date of registration and propensity to attend
There has always been a rather odd relationship between the date someone registers to attend a show (how many months or weeks out from the opening date) and how likely they are to attend. And it is a perverse relationship. Superficially the earlier someone registers, then the more interested in any event they are likely to be and they will have booked out their diary a long way ahead – and thus won’t arrange anything else on those days.
But it doesn’t work that way. The relationship is back to front. The later someone registers, the more likely they are to attend. There are no absolute numbers for every show, but a registrant six months out is less than 20% likely to attend, someone who registers three days before the show, often over 70%. There is, I think, a clear causal negative relationship here – it is not coincidental or random.
So going back to lower conversion rates. The most obvious conclusion is that shows have valid but older databases, with a large percentage having registered a long time ago. The second conclusion is hypothesis, but a large percentage of the workforce is not back in the office full time. Decisions to go to an event are very often made in conjunction and discussion with colleagues. Those discussions are not yet happening as they were by the coffee machine. To some degree, someone who takes time off from “working at home” rather than working in the office, finds the “loss” of that time and the effort of going to a show more significant.
There is still a reluctance by some of the population to travel and attend large gatherings while Covid-19 is still untamed – this is particularly true in the US, where people travel longer distances. Remember it is only five months ago that we were in effective lockdown with Omicron.
I don’t think conversion rates can be considered objectively for at least another year. Only then can we hypothesise about whether, for whatever reason, the propensity to attend events has changed.

Geographical conversion rates
Another interesting – if obvious – factor on conversion rates is geography. I am using data from shows in Birmingham a few years ago, but I cannot see why the conclusion would change. The further away from the location of the show, the less likely the registrant is to attend. In these cases, someone who registered and lived in Plymouth was only 10% likely to attend the show. From Portsmouth only 15%. Someone who registered and lived in Coventry was 60% likely to attend.
At the point of registering, the individual was judging how interesting and relevant the event might be to her/him. Location at that moment was not necessarily critical. But as the date looms, the effort involved in travelling 200 miles, staying overnight etc, became central – and it was easy to decide to give this one a miss. But for someone 10 miles away, it was an easy car or train ride and even a two hour stay at the show was quite possible.
This does have relevance. Some events have a truly national audience (dentists, pharmacists. The Furniture Show, Spring Fair etc). That makes Birmingham a destination where travel considerations are less of a factor than a London event. Conversely, the British transport system is still geared around “all train lines lead to London” which makes it often quicker to get to London from Edinburgh or Newcastle than to Birmingham.
So do make a point of analysing what your pre-registrations are telling you. Create simple doughnuts from the postcode analysis (if you don’t already do it, the software can be downloaded in seconds). You might find that you have a problem you hadn’t recognised.

Which brings us on to correlation, causation and coincidence
We can find confusion between these related terms in our own industry. The most frequent error, much loved by the Daily Mail, is mistaking a correlation for causation. One recurring article is; “Married men are happier than unmarried men” – hence, according to the Mail, marriage makes you happy. Causation errors don’t come any simpler than this. There are two sample groups –married men and unmarried men. The married (or co-habiting) group is much the larger.
But the simple question is: “Why aren’t the smaller group married?” And there are a number of answers. But predictable research shows that men who have always been unhappy or depressed, or men who have mental or physical challenges, or who are socially awkward, are less likely to have found a permanent mate than the “average” man. So the unmarried group is likely (though not universally) to always have been less happy or in some other way more challenged. The institution of marriage is not a causation of happiness, but a correlation of personality (happy and well balanced men are more likely to have married in the first place).
The Mail rarely comments on whether married women are happier.
Spurious correlation occurs when two factors appear causally related to one another (in other words one causes the other) when they are not.
The appearance of a causal relationship is often due to a chance which seems to suggest two things move in tandem – whereas in fact they are simply coincidental or caused by a third factor (your alarm clock goes off around the time the sun rises – but the sun doesn’t cause the alarm to ring, nor does the alarm clock cause the sun to rise).
There are a vast number of similar fun packed examples.
There are two American Football Leagues – the NFL and the AFL. The winner of each has played off for the Superbowl since 1966. For many years, if the NFL team won the Superbowl, then the stock market rose. If the AFL team won, then the stock market would fall. This became so well known, pundits occasionally used it to influence their projections for the markets.
Amazingly this proved true every year from 1966 to 1978. Between 1967 and 1998 it was correct 95% of the time. After that the relationship faded but for the 55 years of the Superbowl it has proven true 75% of the time.
How to explain? Obviously pure coincidence is one possibility. But there is another correlation. The NFL has been the more successful League, especially in the early years, and the stock market has risen annually far more often than fallen. So there is a correlation between two unconnected facts – and presumably no causation.

Shark attacks and ice cream are perfectly correlated
There is a very precise graphical relationship in the US between shark attacks and ice cream sales. So does eating ice cream cause sharks to attack? Of course not. Ice cream sales are much higher in the summer and far more people swim in the ocean in the summer. So there are bound to be more shark attacks than in the winter. A perfect example of a correlation which has no causation.

Or try this one – measles and marriage

There is an almost perfect correlation between measles cases and the rate of marriage in the United States. The graph shows it very clearly.
Both have fallen almost hand in hand for 60 years. So is one related to the other?
Of course not. Measles cases have fallen constantly because of increased vaccination (though this trend has been reversed in the past three years because of the large number of anti-vaxxers – particularly in California).
And, of course, the marriage rate has fallen consistently in the same period as more couples co-habit and, indeed, as the very institution loses the social inevitability of previous generations.
The almost perfectly aligned graphs are simply a coincidence.

What to watch out for in trade show company results
Let’s go back to our own industry and something we will have to watch out for in the next few years –which is how our companies present their results.
There are two graphs below. They use exactly the same information. They are taken from the actual results and current projections of one exhibition/conference company – I won’t identify it.
The first graph – with red bars – shows the audited profit of this company from 2015 to 2021 and then the current projections of the management from 2022 to 2024.

The company had not been doing particularly well from 2015 to 2019, with profits barely improving at all. Then in 2020 Covid hit and profits naturally fell. They are projected to improve through to 2024 – but not to the level they were at in the previous non-Covid year, 2019. In truth, this is not a very good performance. Where have those profits gone? And was the company beginning to spiral downwards anyway in 2019, pandemic or not?

How to produce graphs to fool your shareholders
But look at the graph with green bars. We can imagine the financial director proudly presenting this graph at the Annual Meeting of Shareholders in June 2025 – “Please look at our company’s performance over the past five years and how wonderfully we have recovered from the pandemic. Pay rises all round”

The numbers are identical, but by removing years 2015 to 2019 from the graph shareholders are given a very different impression. If the CEO and FD are relatively new, they can take the credit and ignore the past.
This is about the simplest trick imaginable in presenting annual numbers – there are many more which are a lot more subtle – but in the post-pandemic era a very critical look at what companies say will be no bad thing.
I’ll come back to some of those presentation tricks later.

Related Articles