Enhanced directory listing pricing works when you can show the click difference

An exhibitor emails in March asking whether the 900 dollar enhanced listing is worth it. Somebody in exhibitor services forwards the question around, and the replies are all the same shape. It gets you a logo, a longer description, three product images, and a higher position in category browse.

None of that answers the question. The exhibitor asked what it does, and enhanced directory listing pricing that cannot answer that is asking a buyer to take a view on a product with no stated output.

CEIR's B2B Exhibition Sponsorship Playbook, Part 1, published in October 2019 from more than 200 organiser executives and 728 exhibitors, found that 64 per cent of organisers offer an enhanced listing in the exhibitor directory and 22 per cent of exhibitors buy one. Roughly two thirds of shows sell it and roughly one exhibitor in five takes it, which is the profile of an asset with a real audience and a weak proof.

The number you already have and are not using

Your directory runs on a web platform, and web platforms count. Profile views, outbound clicks to the exhibitor's own site, contact form submissions, brochure downloads, add-to-planner actions. Most organisers have all of this sitting in an analytics property nobody has queried since the platform was configured.

That data is the entire basis for pricing the upgrade. The upgrade is worth the difference in those counts that it causes, per unit, and everything below is about the word "causes".

Start with the naive version, because it is the version most teams reach for and it is worth seeing what it claims. Pull median profile views for every upgraded listing and every standard listing over one edition. Say upgraded listings show a median of 340 views and standard listings 110. The difference is 230 views. The upgrade costs 900 dollars. That is 900 divided by 230, which is 3.91 dollars per extra view.

Three dollars ninety-one is a defensible-sounding number and it is almost certainly wrong.

Why is the gap between upgraded and standard listings not the value of the upgrade?

Exhibitors choose whether to upgrade. That single fact breaks the comparison.

Think about who buys a 900 dollar listing upgrade. Companies with a marketing budget line for the show. Companies taking a larger stand, which means they appear in more places on the floorplan and in more of your own promotion. Companies with an established brand in the category, which people search for by name. Companies with somebody whose job includes filling in exhibitor portals properly, which means their description is longer and their images are better regardless of tier.

Every one of those traits raises profile views on its own. The 230 view gap contains the effect of the upgrade and the effect of being the sort of exhibitor who buys one, mixed together, and nothing in the arithmetic separates them.

Rosenbaum and Rubin set out the standard treatment for this in Biometrika in 1983. Their propensity score is the conditional probability that a unit receives the treatment given its observed characteristics, and they showed that adjusting for that single scalar removes the bias from all the covariates that went into it. The practical version for a directory is less formal and gets most of the way there.

Building a comparison set you can defend

Match before you subtract. For each upgraded listing, find standard listings that look like it on the characteristics that plausibly drive views on their own, and compare inside those matched groups instead of across the whole file.

Four matching variables cover most of it for a trade show directory.

Product category. A listing in a crowded category with 180 companies gets browsed differently from one in a category with nine. Never compare across categories.

Stand size band. Use the contracted net square metres in bands, not the raw number. Stand size is the best single proxy you have for exhibitor scale, and it is on your own contract file.

New or returning. A returning exhibitor carries name recognition and last year's inbound links. A first-timer does not.

Profile completeness. Count the filled fields on the standard listing template only, so you are measuring diligence rather than the upgrade's extra fields.

Now redo the arithmetic inside the matched set. Suppose upgraded listings still show a median of 340 views, but the matched standard listings show 210 rather than the 110 you got from the whole file. The difference is 130 views. The upgrade costs 900 dollars, so 900 divided by 130 is 6.92 dollars per extra view.

That is 77 per cent higher than the naive figure, and it is the number you can put in front of an exhibitor without a competent analyst on their side taking it apart. It is also still an upper bound, because matching only removes the bias from the traits you matched on.

The cleanest test is one you can run next edition

Matching is a repair. Randomisation is a design, and it is available to you in a way it is not available to most people who write about this.

Kohavi, Tang and Xu, in Trustworthy Online Controlled Experiments, published by Cambridge University Press in 2020, make the case that random assignment is what licences a causal reading of a difference, and they draw on organisations each running more than twenty thousand controlled experiments a year. A directory upgrade is an unusually easy thing to randomise, because the treatment is a database flag and the outcome is already logged.

Two designs work in practice.

The first is a comped holdout. Take a group of comparable non-upgraded exhibitors, give the upgrade free to a randomly chosen half for one edition, and measure both halves. You have given away some inventory that would mostly not have sold anyway, and you have bought a causal estimate you can quote for three years.

The second is a randomised offer. Where the upgrade is genuinely scarce, randomise who gets offered it rather than who receives it, and compare offered against not-offered. That estimates the effect of the offer, which is smaller than the effect of the upgrade and is arguably the more useful number, since the offer is what you actually control.

Either way, run it on one category first. A category with sixty exhibitors and a twenty per cent upgrade rate gives you twelve treated listings, which is thin. Two hundred exhibitors across three matched categories is workable.

One warning about the timing. Directory traffic is not flat across the campaign. It climbs from about six weeks out, peaks in show week, and keeps a tail for two or three weeks afterwards as people look up companies they met. An exhibitor who upgraded in January and one who upgraded ten days before doors open have had very different exposure windows, so record the date the upgrade went live and count views only from that date forward. Skipping that step makes early buyers look like the upgrade works and late buyers look like it does not, when the only difference is how long the flag was switched on.

What should the rate card say about the upgrade?

Replace the feature list with a measured claim and a report.

Say what the upgrade produced last edition, in the exhibitor's own category if you can segment that far, expressed as extra profile views and extra outbound clicks against a matched comparison group. Say how the comparison group was built, in one sentence. Give the per-unit cost, so a marketing manager can hold 6.92 dollars a view against whatever their display advertising costs.

Then commit to a post-show report showing that exhibitor's own numbers, with the matched median next to them. An exhibitor who bought the upgrade and got 190 views against a matched median of 210 has a bad result, and telling them so is what makes the good results believable next year. The same discipline is what a dedicated send to the registration list needs on its own delivered and click counts.

Price the upgrade against its measured output, and the awkward pricing question resolves itself. If the matched delta is 130 views and your directory audience is the buyer group the exhibitor wants, 900 dollars may be low. If the matched delta comes back at 18 views, you have found out that your upgrade is mostly cosmetic, which is worth knowing before you print another card. Whether it then belongs in a tier or on its own line is a packaging decision with different economics, and the same output-per-unit logic applies to what you sell inside the event app.

Where this stops

A profile view is a weak outcome and it should be labelled as one.

Somebody scrolling a category page and glancing at nine listings generates nine views. The exhibitor cares about the person who read the description, clicked through and turned up at the stand on Wednesday morning, and your directory analytics cannot see the last step. Outbound clicks are closer to intent than views, and add-to-planner is closer still, but the volumes on those are small enough that medians get noisy fast in a category with sixty exhibitors.

There is also a substitution effect the matched comparison cannot catch. Category browse is a finite surface. If upgraded listings sit higher, they take views from standard listings in the same category, so part of your measured delta is transfer rather than creation. Selling more upgrades in a category shrinks the delta each one produces, and the exhibitors who did not upgrade get quietly worse results than they did the year before. Anyone building this into the wider picture of exhibitor and sponsor performance should keep the cap in view: at a hundred per cent upgrade take-up in a category, the measured value of the upgrade is zero.

This week, export profile views by listing for your last edition, join it to the contract file for category and stand size, and compute two medians inside your largest category: upgraded and standard. Then compute them again restricted to a single stand size band. The distance between those two answers is how much of your current pricing story is self-selection.