Skip to content

Measuring exhibitor driven registrations without taking credit away from marketing

Acquisition and attributionUpdated 2026-08-188 min read

In short

Exhibitor driven registrations should be reported in three buckets: people new to the database, people already known but untouched by your own marketing this cycle, and people who had already registered before the code was redeemed. Only the first two buckets are exhibitor generated, and on many shows they account for under half of all redemptions.

The exhibitor invitation programme closed with 3,400 code redemptions, and the sales director opened the review meeting by saying the exhibitors had brought 3,400 registrations. The audience acquisition lead said quietly that most of those people were already on the file, and had been mailed four times. Neither of them had a number that settled it, so the meeting settled it by seniority.

Exhibitor driven registrations are the most contested figure in an acquisition report, because two teams have a legitimate claim on the same person and only one number gets printed. The argument is worth having properly, once, with a rule that both sides agree to before the codes go out.

What an exhibitor invite code actually proves

A code proves that somebody typed it. That is the whole evidential content of a redemption, and every claim built on top of it is an inference.

CEIR's Attendee Acquisition Trends Driving Growth series, published in 2023 across three reports covering B2B exhibitions held and planned from 2020 through 2023, treats exhibitor and partner invitation programmes as one of the collaborative tactics organisers use to rebuild attendance volume. The series is about which tactics get used and which marketers rate as effective. It cannot tell you whether the person who typed your exhibitor's code was already coming, because that question is answerable only inside your own file.

So the useful work is not defending the programme. It is deciding what the redemption adds on top of what you already had, and that requires knowing what you already had at the moment the code was typed.

Three buckets, and a rule that puts each redemption in exactly one

Every redemption goes into one of three buckets, evaluated in this order.

Already registered. The person completed registration before the code was redeemed. The code changed their badge type or their price, and it did not produce the registration. Test on timestamps: redemption time later than registration completion time.

Known but unengaged this cycle. The person matches an existing record in your database, and in the current cycle they have received nothing from you, or have received campaign email and never opened or clicked it. This is the bucket that decides most arguments, so define it in writing before the campaign and do not move it afterwards.

New to file. No record in the database matches the person at the time of redemption. Nothing you sent could have reached them, because you had no address for them.

The order matters. A person can be new to file and already registered at the same time, if they registered through a different route earlier in the week and only later got the code from an exhibitor. Evaluating registration status first stops the same person counting as an acquisition when they had already acquired themselves.

Bucket two and bucket three both depend on matching a redemption to a record, which is a probabilistic problem rather than a lookup. Fellegi and Sunter set out the standard treatment in the Journal of the American Statistical Association in 1969: you classify pairs of records, not records, and a pair falls into match, non-match, or a middle band a human reviews. Run this with an exact email join alone and your new to file bucket inflates, because a person who registered in 2024 from a work address and redeemed in 2026 from a personal one looks new.

Working the split on 3,400 redemptions

Take the 3,400 redemptions and apply the three tests in order.

Registration completed before redemption: 640 records, which is 18.8 per cent of the file. These are price changes, and the honest label is exhibitor assisted rather than exhibitor generated. Set them aside.

That leaves 2,760. Match those against the database as it stood on the redemption date. Suppose 1,910 match an existing record and 850 do not. The 850 are new to file.

Now split the 1,910. Of those, 1,180 had at least one campaign email opened or clicked in the current cycle, or arrived on a landing page from a tagged paid click in the 30 days before redeeming. They were in your funnel, and the exhibitor's invitation was one touch among several. The remaining 730 had nothing: no open, no click, no tagged session.

Exhibitor generated is the new to file bucket plus the known but unengaged bucket, so 850 plus 730, which is 1,580. Against 3,400 redemptions that is 46.5 per cent.

The number the sales director wanted was 3,400. The number the acquisition lead implied was closer to 850. Both were wrong, and the gap between them is 730 people your marketing had on file and had failed to reach, whom an exhibitor reached instead. That is a real contribution and it is worth paying for.

Which bucket counts as exhibitor generated?

I would count buckets two and three, and I would fight for bucket two rather than conceding it.

The argument against bucket two is that these people were in your database, so you could have reached them. You did not. A record you own and never activate has produced nothing, and the person who activated it has done work you did not do. If your position is that anyone on file belongs to marketing forever, then every channel you ever run is uncredited after the first edition. That position claims ownership of a record. It says nothing about what caused the registration.

The argument for keeping bucket two conservative is stronger than it looks though. Unengaged is a threshold you set, and the number moves a long way depending on where you put it. Defining unengaged as no open in the cycle produced 730 above. Define it as no email delivered in the cycle and the bucket shrinks to whatever share of your file you genuinely never mailed, which on most shows is small. Define it as no click and the bucket grows, because open rates are unreliable evidence of anything.

Pick the click definition or the delivery definition, write it into the exhibitor programme documentation, and publish the count under both if the exhibitor asks. The definition costs you nothing to disclose and it removes the suspicion that the rule was chosen after the result was known.

What do you tell the exhibitor whose code mostly reached your own database?

Tell them the split, with their own numbers in it.

An exhibitor who issued 400 invitations and drove 180 redemptions, of which 34 were new to file and 41 were known but unengaged, has delivered 75 registrations that would probably not have happened. That is a good result for a single stand, and the 105 in the other buckets are still evidence that their customers came to see them, which is a different and also useful thing to know.

What this rules out is the leaderboard that ranks exhibitors on raw redemptions. A large exhibitor with a 60,000 record customer list will top that table every time by mailing people who are already yours, and a small specialist who brought 20 genuinely new buyers will sit near the bottom. Rank on the exhibitor generated count instead, and the ranking starts rewarding the behaviour you actually want. The per code arithmetic for individual named advocates, where the same skew shows up much harder, is B33's ground, and the ranking of speaker and ambassador codes follows different rules because those codes are issued to people rather than companies.

Where this stops

The three buckets rest on two measurements that are weaker than they look.

The first is the match. Every record that fails to match becomes new to file by default, so your exhibitor generated count is inflated by exactly your matching error rate, and the error is not random. It concentrates in people who changed employer, which is the population most likely to be a genuinely valuable buyer and also the population most likely to look new when they are not. Report the new to file bucket with the match rate printed next to it, and rerun the split with a looser matching threshold to see how much the answer moves. If moving from an exact email join to name plus company matching drops your new to file count by a third, that is the size of the uncertainty.

The second is that unengaged is a statement about your systems. Somebody who deleted three of your emails unopened has been reached, in the ordinary sense of the word, and your data says they were not. This cuts against my own recommendation above, and I still prefer the click definition, because the alternative is treating a delivered email as a claim on a person indefinitely.

There is also a limit that no rule fixes. An exhibitor who tells a customer to come, and the customer then searches the show name and registers without the code, produces nothing you can see. Your exhibitor generated number is a floor, and everybody in the room should know that before they use it to price a sponsorship.

This week, pull last edition's code redemptions with their timestamps, join them to the registration table, and count how many redemptions happened after the registration was already complete. That single number, the size of the first bucket, usually settles more of the argument than the rest of the analysis put together, and it takes one query against data you already hold. From there, the acquisition and attribution reporting can carry all three buckets rather than one contested total, and the same discipline extends to paid partners once money is attached to the claim, which is where commission rules for affiliates come in and where the cost of a qualified registration decides whether the programme was worth running at all.

Questions people ask about exhibitor driven registrations

What counts as an exhibitor driven registration?
A registration where the exhibitor's invitation brought a person your own marketing had not already brought. That means either a person with no prior record in your database, or a person on file who had received nothing and opened nothing in the current campaign cycle. A redemption from somebody who registered last week is a code applied to an existing registration.
How do you stop exhibitor codes taking credit away from marketing?
Compare the redemption timestamp against the registration timestamp and against your own campaign history for the same person. A code redeemed after the registration was already complete cannot have caused it. A code redeemed by somebody who clicked your campaign email nine days earlier is shared credit, and reporting it as exhibitor generated moves the same person twice.
Why are so many exhibitor code redemptions people already in the database?
Exhibitors invite the customers and prospects they already have, and on a mature show a large share of those people are on your file too. The overlap is not a failure of the programme. It is a measurement problem, and the fix is to size the overlap before the campaign rather than argue about it afterwards.