Estimating booth price elasticity when you only have five editions of data

Your finance lead asks what happens to volume if the premium zone goes up eight per cent. That is a question about booth price elasticity, it has a number as an answer, and so you go and look for the number.

What you find is five rows. Five editions, five rate cards, five totals for net square feet sold. Two of those editions were in a different hall, one moved from March to June, and 2021 should probably be thrown out entirely. You cannot fit a demand curve to that and you know it, so the answer that goes back is that it depends, and the rate card gets set the way it was set last year, which is last year plus a bit.

There is a better answer available, and it does not come out of your historical file at all.

Why will five editions not give you a demand curve?

The reason is not sample size on its own. It is that price and quantity moved together for reasons that had nothing to do with each other.

In 2022 you raised the rate four per cent and sold more space, because the category was recovering. In 2024 you held the rate flat and sold less, because two of your largest accounts merged and took one stand instead of two. If you regress net square feet on rate across those five points you will get a coefficient, and it will be a summary of your category cycle wearing a pricing coefficient's clothes.

There is a second problem underneath that one. Booth space is not a single product with a single price. A 100 square foot inline at the back of hall B and a 900 square foot island on the main cross aisle are different goods sold to different buyers at different rates, and your five annual observations are an average over a mix that changed every year. Even with twenty editions you would be estimating one elasticity for a basket, which is not the number anybody actually wants when they ask about the premium zone.

The context makes the question harder to dodge. The Center for Exhibition Industry Research reported that in the second quarter of 2025 the CEIR Total Index sat 8.4 per cent below the same quarter of 2019, with net square feet down 4.9 per cent and real revenues down 15.6 per cent against that 2019 base. Space volume has recovered further than inflation adjusted revenue has. Whatever has been happening to realised rate, it has been working against you, and guessing at the elasticity is how organisers end up there.

Ask the buyer instead, the way Gabor and Granger did

Gabor and Granger published their approach in Economica in 1966, in a paper on price as an indicator of quality, and the mechanic is almost rude in its simplicity. You name a price. The respondent says whether they would buy at that price. If they say yes, you name a higher one. If they say no, you name a lower one, and you keep going until you have found the point where each individual buyer switches.

Aggregate those switch points across respondents and you get a buy response curve: the proportion who would accept at each rate. Multiply the proportion by the rate and you have a revenue curve, which is the thing you were trying to estimate from five rows of history.

This transfers to exhibition space better than it transfers to most products, for two reasons. Your buyers are professionals who have a budget line for this and can answer the question honestly. And you already have the conversation scheduled, because renewal calls happen anyway, so the marginal cost of asking is a field in the call script.

The four questions from 1976

Van Westendorp presented the Price Sensitivity Meter at the twenty-ninth ESOMAR Congress in Venice in September 1976. It asks four questions instead of one ladder.

At what rate would this be so expensive you would not consider it. At what rate would it start to feel expensive, so that you would think hard before signing. At what rate would it feel like good value. At what rate would it be so cheap you would wonder what was wrong with the show.

Each question produces a cumulative distribution across respondents, and the crossing points of those four curves give you a band. The crossing of too cheap against expensive marks the bottom of the acceptable range. The crossing of good value against too expensive marks the top. What you get is a corridor. The four curves cannot produce a single optimum, and the corridor is often the more useful object anyway, because the published rate card has to hold for a full selling season.

The fourth question is the one organisers want to cut, and it is the one worth keeping. Exhibition space carries a quality signal. An exhibitor who sees a premium island offered at a rate close to an inline reads that as a show in trouble, and the too cheap curve is where you find out at what point that starts happening.

Working the ladder on sixty renewing exhibitors

Take a premium zone currently at $34 per square foot and a standard 10 by 20 inline, so 200 square feet at $6,800. Put the ladder in front of sixty renewing exhibitors during the renewal call and record acceptance at each rate.

Say the answers come back like this. At $32, 57 of the 60 accept. At $34, 54. At $36, 52. At $38, 46. At $40, 35. At $42, 22.

Now do the revenue arithmetic, holding the booth at 200 square feet. At $32 the booth is $6,400 and 57 buyers gives $364,800. At $34 the booth is $6,800 and 54 buyers gives $367,200. At $36 the booth is $7,200 and 52 buyers gives $374,400. At $38 the booth is $7,600 and 46 buyers gives $349,600. At $40, 35 buyers at $8,000 is $280,000.

The peak sits at $36, and the fall away from it is asymmetric. Going one step below the peak costs you $7,200. Going one step above costs you $24,800. That asymmetry is the single most useful output of the exercise and it never appears in a percentage increase conversation.

How do you read the elasticity number?

Elasticity is the percentage change in quantity divided by the percentage change in price, and with a buy response table you can compute it between any two rungs.

From $34 to $36, acceptance falls from 54 to 52. That is a fall of 2 on a base of 54, or 3.7 per cent. The rate rises by $2 on a base of $34, or 5.9 per cent. Divide 3.7 by 5.9 and you get 0.63. Below one, which means demand is inelastic over that step, which means the price rise more than pays for the volume it costs. That is the licence to move.

From $36 to $38, acceptance falls from 52 to 46. That is 6 on a base of 52, or 11.5 per cent, against a price rise of 2 on 36, or 5.6 per cent. Divide 11.5 by 5.6 and you get 2.07. Above one. Every dollar you add past $36 is now costing you more in lost booths than it earns on the ones you keep.

Go back to the opening question. Eight per cent on $34 is $36.72, which lands just past the point where the elasticity crosses one. The honest answer to your finance lead is that six per cent is defensible and eight is on the wrong side of the break, and you can show the table the answer came from. Where that leaves the size of the increase you publish is a separate decision with its own constraints.

Running both, and what to do when they disagree

Run the ladder and the four questions on the same sixty people in the same conversation. Together they take about four minutes.

Suppose the four questions produce a lower bound near $28, an upper bound near $41, and an indifference point near $35. The revenue peak from the ladder was $36. That agreement is the result you want, and it means you can defend $36 on two independent pieces of evidence.

Disagreement is more interesting. If the ladder peaks at $36 but the too expensive curve says a third of your renewers are out by $34, you are looking at a segment split rather than a measurement error, and the fix is a zone line rather than a single number. If the ladder peaks above the upper bound of the acceptable band, someone has anchored the respondents, which brings us to the part where this method misleads you.

Where this stops

Stated intention is not a signature, and it never becomes one.

The specific failure with the Gabor and Granger ladder is anchoring. The first rate you name pulls the answers toward it, so the starting rung is itself an input to the result. If your account manager opens at $40 with one exhibitor and $32 with another, you have measured your account managers. The mitigation is to randomise the starting rung across respondents and to have the ladder read out by someone who is not carrying that account's number, and to accept that the level is less trustworthy than the shape.

The second failure is who answers. You are asking renewing exhibitors, which means everyone in your sample already decided your show is worth attending. The exhibitors whose elasticity you most need to know are the ones who dropped out after 2023, and they are not on the call. A buy response curve built on renewers will always look more inelastic than the market is.

The third is that neither method knows anything about your competitors. Both were designed for a product sitting on a shelf next to alternatives the respondent can see. An exhibitor deciding between your March show and a rival's May show is making a comparison your survey never mentioned, and the acceptance rates you collect quietly assume that comparison came out your way.

None of this makes the exercise worthless. It makes the output a band with a shape, which is a great deal more than five rows of history will ever give you, and it is checkable against what actually renews. What you then do with that band once the season is running, and whether the rate moves inside it, is a separate mechanic with its own failure modes.

Before the rate card locks, pick the sixty largest renewing accounts in one zone, add six lines to the renewal call script with the ladder rungs randomised, and put the answers in a field on the account record, which is the form exhibitor analytics needs them in, so that next year you can compare what they said to what they signed.