Writing the post event report executive summary a chief executive will actually read

The chief executive of a mid-cap exhibitions group told me she reads the first page of a post-show report and nothing else, and that she was not embarrassed about it. She has eleven shows. Eleven reports at forty pages is four hundred and forty pages a year, and she has a portfolio to run.

Her complaint was specific. The first page was always a paragraph of prose followed by a chart, and by the time she had worked out what the chart's axis was measuring she had lost interest in the paragraph. She wanted five numbers and a sentence each.

That is a completely reasonable request and most organisations cannot fulfil it, because the summary page is written last, by whoever is left, from whatever the deck already contained. The page sits on top of a much longer document, and what belongs in that report and what gets cut should be settled before anybody writes the first line of the summary.

How many numbers belong on the summary page?

The summary page carries five metrics. Five is enough to describe a show as a business and few enough that a reader holds all of them at once, which is the whole of the argument for that number.

Each metric gets four things: this edition's figure, the prior edition's figure, the absolute change, and one sentence saying what caused it. The percentage goes next to the two figures it came from, so the division is checkable without turning a page.

Choose the five once, for the series, and keep them. The temptation every year is to swap in whichever metric flatters this edition, and giving in to it once destroys the page's usefulness permanently, because a reader who has learned that the five rotate stops treating any of them as a trend. That is the same discipline that holds the rest of the pack together, and standardizing show reports works it through at the level of the whole document.

Rank the five by how unusual each move is

The usual ordering rule is percentage change, largest first. It is a bad rule, because percentage change is a function of how volatile the metric normally is, and volatility differs enormously between metrics on the same show.

Attendance on a large show moves by a couple of thousand for reasons nobody controls: weather, a competing event, a public holiday landing badly. Exhibiting company count barely moves at all, because it is a book of contracts built over eighteen months. A 3 per cent swing in the first is a Tuesday. A 3 per cent swing in the second is a problem with a name.

So compute, for each of the five, the absolute change this edition divided by the standard deviation of that metric's year-on-year change across your last five or six editions. You need six editions of history to have five changes, which most shows have. The result is a unitless score, and it ranks moves by how far outside normal they sit.

This takes about twenty minutes in a spreadsheet, once, and then it is a formula.

The worked example

A show with 208,400 net square feet of paid space, which puts it above the 200,000 threshold CEIR uses to define a large exhibition in the second edition of its Performance Benchmark Playbook, published in 2026.

Gross revenue is 13,140,000 against 12,760,000 last edition, up 380,000, up 3.0 per cent. Attendance is 41,800 against 43,050, down 1,250, down 2.9 per cent. Exhibiting companies number 1,415 against 1,462, down 47, down 3.2 per cent. Net square feet of paid space is 208,400 against 205,900, up 2,500, up 1.2 per cent. International attendance is 8,360 against 7,320, up 1,040, up 14.2 per cent, taking the international share from 17.0 to 20.0 per cent.

By percentage change, the order is international attendance, exhibiting companies, revenue, attendance, space. By the ranking rule above it comes out differently.

The standard deviation of year-on-year change across the last five changes is 210,000 for revenue, 1,900 for attendance, 22 for exhibiting companies, 4,100 for net square feet, and 480 for international attendance. Divide each move by its own figure. Revenue scores 380,000 over 210,000, which is 1.81. Attendance scores 1,250 over 1,900, which is 0.66. Exhibiting companies score 47 over 22, which is 2.14. Space scores 2,500 over 4,100, which is 0.61. International attendance scores 1,040 over 480, which is 2.17.

The summary now leads on international attendance at 2.17 and exhibiting companies at minus 2.14, with revenue third at 1.81. Attendance, the number that will dominate every conversation in the building for the next month, is the least unusual move on the page at 0.66, well inside the range this show produces every year without anything happening.

That reordering is the entire value of the page. Two things genuinely changed on this show, and neither of them is the thing everyone is talking about.

One cause per number, with an honest grade

Each of the five gets one sentence naming a cause, and the sentence carries a grade for how well the cause is evidenced. The causes come out of the attendee analytics sitting under the page, and they vary a lot in how well established they are. I use three words: measured, inferred, and unknown.

Measured means you can point at the data that establishes it. International attendance rose 1,040 and 780 of that increase came from three countries where a new reseller programme ran, which is visible in the registration source field. That is measured.

Inferred means the cause is plausible and consistent with the data but not established by it. Exhibiting company count fell 47 and 31 of those were accounts under 200 square feet in one product category that consolidated during the year, so the fall looks like category consolidation rather than dissatisfaction. That is inferred, and saying so protects you when somebody checks. Decomposing a space move down to accounts and categories is a separate exercise with its own method, and it does not belong on this page.

Unknown means you do not know. Write unknown. A summary page with one unknown on it is more credible than a page with five confident causes, and the unknown is the line that gets somebody assigned to find out before the next edition.

The worst outcome is a cause invented under deadline that then becomes the organisation's settled explanation. I have watched a made-up reason for an attendance dip survive three years and shape a marketing budget, because it was on the summary page and nobody remembered it had been a guess.

The benchmark line, used once and carefully

Chief executives compare. If you do not give them an external reference, they will find one, and it will be a competitor's press release.

One line at the foot of the page is the right dose. CEIR's Performance Benchmark Playbook, second edition, published in 2026, covers business-to-business exhibitions with 200,000 net square feet or more of paid exhibit space and reports median gross revenue of 12.5 million dollars for that group, with 80 per cent of organisers reporting profitability and an average net profit margin of 55 per cent. This show's 13,140,000 sits above that median, which is a fact worth one sentence and no more.

Say what the benchmark's population is in the same sentence as the number. A median across large United States shows is the wrong comparator for a 60,000 square foot regional event, and the way that error usually enters a board pack is that somebody quoted the figure without the population attached. P35 covers reading the CEIR Index from outside the United States and P19 covers reading a benchmark's sample definition, which is where the care belongs.

What a chief executive is worried about is generally not your show in isolation. UFI's Global Exhibition Barometer, 36th edition, published in January 2026 from 378 companies across 57 countries, has respondents naming the state of the economy in their home market as the leading business issue at 19 per cent, ahead of global economic developments and geopolitical challenges at 16 per cent each. A summary page that lands a show's performance against that backdrop gets read differently from one that presents the show as a closed system.

What stays off the page?

No chart. A chart on a summary page costs a reader ten seconds of axis-reading before it tells them anything, and five numbers with prior-year figures already contain everything a chart of five numbers could show.

No commentary on next year. The summary reports the edition that closed. Forecast and plan are a separate document with a separate approval, and mixing them means the reader cannot tell which of the numbers on the page happened.

No metric that appears nowhere else in the report. Every figure on the summary must trace to a section that derives it, and the section number goes next to it in small type. That reference is what converts the page from a claim into an entry point.

One page can serve more than one reader. The same five numbers can sit at the front of every version of the report, which is the cheapest way to get four readers with four different questions starting from the same figures before each of them turns to a different section.

Where this stops

The ranking rule needs six editions of stable history, which rules it out for a show that launched three years ago, moved venue, or changed its definition of attendance in the middle of the series. With three data points the standard deviation is noise and the score will confidently tell you that everything is unusual.

For a young show, fall back to ranking by absolute money where a money figure exists and by absolute count where it does not, and say in the page's footer that the ranking is by size rather than by deviation. That is weaker and it is honest.

The deeper limit is that a summary page cannot fix a disagreement about what a number means. If sales and operations are carrying different attendance figures into the meeting, a beautifully ordered summary page just picks a side, quietly, in front of the board. Settle that first, which D7 covers, then write the page.

Pull your last six editions of one show, compute the year-on-year change in each of your five headline metrics, and take the standard deviation of those five changes per metric. You will have the denominators you need for every future summary page in an afternoon, and you will probably discover that one of your five metrics has never meaningfully moved and does not deserve its slot.