The buying committee at trade shows rarely arrives on one badge

An exhibitor came off a stand on the second afternoon annoyed that the hosted meeting they had waited three months for was with somebody who could not buy anything. The buyer had been polite, taken two brochures, photographed the spec sheet and left. In the exhibitor's account records, that buyer's company had four other named contacts, and none of them had registered.

The meeting was not wasted, and the scoring was still wrong. A buying committee at trade shows sends a scout, and a score that reads only the badge in front of it treats that scout as a junior browser.

Who is actually buying?

A group, most of the time, and one that has been described in the same terms for fifty years.

Webster and Wind set out a general model of organisational buying behaviour in the Journal of Marketing in April 1972, and the piece of it that matters here is the buying centre: purchase decisions in an organisation involve many people, several goals and criteria that can conflict with each other, which makes the decision a process across a group instead of an act by an individual. Bonoma made the practical point in Harvard Business Review in May 1982, arguing that power in a buying decision does not track organisational rank, and that sellers who target the highest title often address somebody with no influence over the choice.

Both papers were written about field sales, and a trade show floor makes their problem sharper. The seller gets 20 minutes with one person, chosen by the buyer's organisation for reasons the seller cannot see. Sometimes that person is the decider. Often they are the specifier, or the gatekeeper, or the one who had the week free.

There is no way to observe the committee directly. There are two things you can observe: what the person declared about their own role, which buyer authority scoring in I14 turns into a number, and what the rest of the account looks like in your own file.

What the account record can tell you that the badge cannot

Take every registration for this edition and resolve it to a company account. Then look at what else that account has done across your portfolio.

Four facts are usually available and rarely used. How many people from the account registered this edition and at what seniority levels. Whether anyone from the account attended earlier editions of this show. Whether the account appears in any exhibitor's lead capture history from a previous year, if you run a lead retrieval product. And how many of your other portfolio events the account shows up at, which is a size and engagement signal that has nothing to do with any single badge.

Resolving registrations to accounts is its own discipline with its own failure modes, and the difference between resolving people and resolving companies is a separate problem that J40 covers. A rough version is enough to start: normalised domain from the work email, with a fallback to normalised company name, and a manual list for the twenty largest accounts where subsidiaries and spellings multiply.

Counting each account once, however many badges it holds

The obvious account measure is the wrong one. Counting registrations per account makes a company that sent six engineers look more powerful than a company that sent one managing director, and the six engineers are a training trip.

Build the account authority from the levels, not the volume. Take the highest personal authority score in the account, then add a bounded increment for each additional person at level 3 or above, and cap the total at 1.0.

Work it through. An account sends a level 2 director scoring 0.80, a level 4 manager scoring 0.45 and four practitioners scoring 0.15 each. The base is the maximum, 0.80. One additional person sits at level 3 or above, which is nobody here since the manager is level 4, so no increment applies and the account scores 0.80. A second account sends two directors at 0.80 and 0.72 plus one head at 0.55. Its base is 0.80, and two colleagues qualify for the increment at 0.05 each, giving 0.90. A third account sends six practitioners at 0.15, and it scores 0.15.

Six badges, one account, one score of 0.15. That is the behaviour you want, and it is the opposite of what a count of registrations produces.

Blending the account term into the personal score

Now put the two numbers together for each individual. Take the personal authority score, and blend the account authority in at 40 per cent.

The scout from the opening paragraph is a level 4 manager with a personal authority of 0.45, from an account whose authority is 0.90. Their blended authority is 0.6 times 0.45, plus 0.4 times 0.90, which is 0.27 plus 0.36, or 0.63.

An isolated buyer with the same title from an account with nobody else registered has an account authority equal to their own, 0.45, so their blend is 0.6 times 0.45 plus 0.4 times 0.45, which is 0.45. Same badge, same parsed title, and a difference of 0.18 in the authority term.

Feed that into the weighted score. Authority fit carries 0.15, so the difference in the total is 0.15 times 0.18, which is 0.027. That is a modest move, and it should be modest, because the account evidence is circumstantial. On a ranked list where the top twenty candidates for an exhibitor sit inside a range of about 0.08, a move of 0.027 reorders things without overturning them.

Blending at 40 per cent is a policy choice, and the parsed seniority level that I15 produces is the input the whole arrangement rests on. Write the 40 down, and test it the same way you would test any other weight, by holding out a prior edition and comparing acceptance across blends of 0, 20 and 40 per cent.

Why does this change which meetings you propose?

Because it changes both the ranking and the allocation, and the allocation is where the money is.

Ranking first. Two buyers with identical titles and identical category interest now separate on whether their colleagues are in the building, and the one representing a live committee moves up. That is the correct direction, and it is invisible to any score that reads one row at a time.

Allocation second. Once accounts exist, you can see a failure that person level matching cannot. An exhibitor with 12 meeting slots across two days receives proposals for three people from the same retailer. Accepting all three spends 25 per cent of their programme on one company, and the retailer sends three colleagues who each hear the same pitch. Deduplicate at account level, propose the strongest contact, and route the other two to exhibitors they have not otherwise met.

The exception is worth building in from the start. Some exhibitors, particularly in capital equipment, actively want the engineer and the buyer in the same conversation, and a few will ask for it. A per-exhibitor flag for account level grouping handles that in a line of configuration, and it turns a scoring rule into something your commercial team can sell.

Accounts also give you something to compare against the exhibitor's own list. Most exhibitors above a certain size arrive with 40 or 50 target accounts they care about, and a fair number will hand that list over if you ask during onboarding. Matching their targets against your resolved accounts produces a short, unambiguous report: of 48 target accounts, 19 have somebody registered, 6 of those have a contact at level 3 or above, and 3 already have a meeting proposed. That report is more useful to an exhibitor than any score, and it exposes the accounts with nobody registered at all, which is a recruitment job for audience acquisition and not a scoring problem. It also gives the account resolution work an owner outside the data team, since exhibitors correct their own account names quickly when the output is a list they recognise.

Where this stops

The committee members who never register are invisible, and no amount of account modelling conjures them. If a buyer's four colleagues have never touched your platform, your account authority is built from one badge and it equals the personal score, which is the same position you started in.

Account resolution errors also propagate in an ugly direction. Merge two unrelated firms with similar names and you create a phantom committee that lifts every person in it. The safeguard is the same one that applies to any merge decision: keep the account assignment reversible, log which rule made it, and hold the largest accounts under manual review, because a false merge at the top of your file moves hundreds of proposals.

The last limit is a fairness one that shows up in the numbers a year later. Large accounts get more badges, and more badges raise account authority, so the blended score systematically favours enterprises over the 20 person specialist that sends its owner alone. If your show's value proposition includes access to small innovative suppliers, cap the account increment low, as in the 0.05 above, and watch the share of proposals going to the top 50 accounts each edition.

Start by counting badges per account for last edition. One query, grouped by resolved account, with the parsed seniority levels alongside. If the median account has one registration and the top decile has five or more, you already know which proposals your current score is underrating, and the account term is the cheapest correction available to your matchmaking configuration.