Corporate email domain share as an early read on audience quality

A paid burst goes live on a Monday. By Friday the weekly registration number is up 3.5 times on the week before, the show director is pleased, and the audience acquisition lead has a slightly uneasy feeling she cannot yet defend with a number.

Nine weeks later the exhibitor advisory board asks who all these people were. By then the answer needs badge scan data, a post-show survey and a fortnight of work, and the money is spent either way.

There is a cheap read available on the Friday, and it is corporate email domain share. Count the share of that week's registrations whose email address sits on a domain the registrant's employer owns instead of a free consumer provider. It is one derived column, it costs nothing to compute, and it moves fast enough to catch a channel drifting inside a single week.

What does corporate email domain share actually measure?

A corporate domain is weak evidence of one specific thing: that the person filled in the form in a work context, using the identity their employer gave them. That is all it tells you. It does not say they are senior, it does not say they buy, and on plenty of legitimate audiences it says nothing at all.

What makes it useful anyway is that it is available immediately, applies to every record, and needs no self-reported field. Every other quality read you have is either slow, partial or answered by the registrant about themselves.

The industry's own quality measurement is slow by construction. UFI and Explori presented their Global Visitor Insights study in December 2018, drawing on more than 13,000 visitor responses from 135 countries alongside in-depth interviews with 29 event directors and senior marketers. It found overall satisfaction at 3.86 out of 5.00 with a net promoter score of plus 7, and it found that between 22 and 27 per cent of visitors in mature markets felt the shows they attended were getting worse. Those are careful numbers and they arrive after the show. A weekly domain share arrives while the campaign is still running, and the two answer different questions. One tells you what your audience thought. The other tells you which audience the campaign is currently buying. Neither is the weighted score you would use to rank individual registrants, which needs several fields and a published lookup table.

Defining the free list so the series is comparable

The number is only as stable as the list behind it, and the list is where most of the argument lives.

Start with the obvious consumer providers and their national equivalents, which is the part everybody gets right and then stops. A serious list also has to handle four other categories.

Internet service provider addresses. Domains handed out with a broadband connection. They are not free webmail and they are not corporate, and in some markets they are a large share of the file.

Educational and government domains. Owned by the employer, so they pass a naive corporate test, and on most B2B shows they are a distinct audience with distinct behaviour. Split them out rather than folding them into corporate.

Disposable and throwaway domains. These are a different problem from consumer webmail. Castle published an open source list in May 2026 covering the top thousand disposable email domains it observed in real abuse activity, updated daily, and deliberately excluded privacy forwarding services such as SimpleLogin and Addy on the grounds that their purpose is privacy. That distinction is the one to copy. A throwaway inbox says nothing about audience quality and quite a lot about whether the record is real, which belongs to the screen for fabricated registrations.

Custom domains on hosted mail. A twelve person distributor running its mail through a hosted suite on its own domain is corporate. The domain is not on anyone's free list and never will be, which is fine, and it is worth remembering when someone proposes recognising corporate domains from a whitelist instead of a blocklist.

Version the list with a date, and when you add entries, backfill the historic weeks or accept that the series has a step in it. A domain share that moves because the list moved is the most common way this metric gets discredited.

Cohort by acquisition week, never cumulatively

The cumulative share across the whole file is nearly useless as a monitoring number, because the file is large and the new arrivals are small, so the cumulative figure moves slowly no matter what is happening.

Work an example. By week 15 of the campaign the file holds 8,000 registrations at a corporate domain share of 77.0 per cent, which is 6,160 corporate addresses. The paid burst then lands.

Week 16 brings 1,880 registrations, of which 1,222 are corporate, a share of 65.0 per cent. Week 17 brings 2,240 registrations, of which 1,366 are corporate, a share of 61.0 per cent. Week 18, after the burst is turned off, brings 900 registrations with 660 corporate, back up to 73.3 per cent.

Watch the two views diverge. The weekly cohort fell from 77.0 per cent to 61.0 per cent, sixteen points, and recovered most of the way in one week. The cumulative view adds 5,020 registrations at a blended 64.7 per cent to a base of 8,000 at 77.0 per cent, giving 9,408 corporate addresses on 13,020 registrations, which is 72.3 per cent. The cumulative number fell 4.7 points and it fell smoothly, which is exactly the shape that gets read as gentle drift rather than as a channel doing something specific in a specific fortnight.

Report the weekly cohort. Keep the cumulative figure on the same chart as a flat reference line, since it is what the file looks like in aggregate and somebody will ask, but the decision is made on the cohort.

Reading it against the channel that produced it

The cohort number tells you something happened. Splitting it by acquisition source tells you where.

Add the source to the same weekly cut. In the example above, if organic and email registrations held at 79 per cent through weeks 16 and 17 while the paid social cohort ran at 44 per cent, the burst did not degrade your audience. It added a second audience alongside the first one, and the blended number is an average of two populations that should never have been averaged.

That distinction changes the decision. A channel running at 44 per cent corporate is not automatically wrong. If your show has a legitimate sole trader and small operator audience, those people genuinely use consumer addresses, and a campaign that reaches them is doing its job. What you cannot do is buy that audience while reporting a headline registration number to exhibitors who were sold something else.

The useful posture is to hold a floor per channel rather than a floor for the show. Agree in advance what corporate share you expect from each source, write it into the media plan, and treat a breach as a conversation about targeting rather than as a verdict.

What should a falling share change?

Two weeks of a channel at 44 per cent against an expected 70 is enough to change the targeting, reallocate the next flight, or ask the agency for the placement report. It is not enough to declare the registrations worthless.

I would go further than most teams are comfortable with and say the number should have a standing consequence attached before the campaign starts. Without one, the chart gets shown, everyone agrees it is interesting, and the burst runs to completion because the registration total is up and stopping it is awkward.

The consequence I would write is narrow. If a paid channel's weekly corporate share sits more than fifteen points below its agreed floor for two consecutive weeks, the next flight for that channel is held until the targeting is changed. That is a small enough rule to survive a real campaign and specific enough that nobody has to relitigate it at the time.

Where this reads badly

Corporate domain share is a proxy, and it breaks in markets rather than at the edges.

In several large economies, consumer webmail is the normal business address for small and medium firms, including firms with real buying budgets. A show with a significant audience in those markets will read a low corporate share as poor quality when it is measuring local practice. If your file has a meaningful international tail, compute the share by country before you compute it for the show, or you will spend a season trying to fix a number that is describing somebody else's mail habits.

Mobile registration pushes the same way. Somebody registering from a phone on a train uses the address their phone is already signed into, and that is the personal one. So the share is partly measuring device mix and time of day, which the paid social burst also changes, which means some of the sixteen point fall in the example above is a mechanical consequence of where the traffic came from rather than a fact about who the people are.

The last limit is the one to say out loud to anyone who wants to promote this into a target. Corporate domain share is not a qualification rule and cannot stand in for one. It reads the surface of an identity. Whether the person specifies, recommends or approves a purchase in a sector your exhibitors sell to is a separate question, and it is answered by the qualification criteria you sold exhibitors on. Keeping the two apart is basic hygiene in any attendee analytics pack.

Start this week by adding one column to the registration export: free, corporate, education or government, and disposable, derived from the domain. Then plot the corporate share by acquisition week for the last completed edition, with the campaign send dates and paid flight dates marked on the same axis. If the dips line up with your own media calendar, you have a working leading indicator before you have written a line of new code.