Converting net square metres to square feet without breaking your benchmark

A portfolio analyst builds a five year space series for a European show so it can be compared against United States sector data. The show's certificates are in square metres. The comparison set is in square feet. Converting net square metres to square feet looks like the least interesting step in the whole exercise, so it gets done in a spreadsheet cell with a factor somebody remembered.

Two weeks later the show appears to have grown 9.5 per cent in a year when it grew 1.8. Nothing was falsified. A rounding rule was applied inconsistently across two editions, and it moved the number by five times more than the market did.

What is the exact factor, and why does it matter to four places?

The square foot has an exact definition in metric terms, inherited from the international yard and pound agreement. The United States National Institute of Standards and Technology publishes it in Appendix B of its guide for the use of the International System of Units, where the factor from square feet to square metres is given as 9.290304 times ten to the minus two and marked as exact (NIST, 2026).

Invert it and one square metre is 10.7639104 square feet. Four decimal places, 10.7639, is plenty for exhibition work. Two, 10.76, introduces about 360 parts per million, which on a 50,000 square metre venue is under 200 square feet and nobody will ever notice.

The rounding that does damage is the one to a single significant figure. Multiplying by ten is 7.1 per cent low, every time, in the same direction. That error is larger than the year on year change in rented space that most shows report, which is exactly the property that makes it dangerous. A systematic bias smaller than the signal is an annoyance. A systematic bias several times the signal is a fabricated trend.

Which unit is authoritative depends on the document. UFI's Auditing Rules for the Statistics of UFI Approved Events, dated June 2021, certify net exhibition space alongside exhibitor counts in square metres. The CEIR Index, published through IAEE, builds its measure from four components of which the first is NSF, net square feet (IAEE, 2026). Both are correct in their own domain, and a portfolio spanning both has to hold the two side by side.

The ten times shortcut and what it costs

Run the arithmetic on a single figure first.

A show certifies 9,000 square metres of net exhibition space. Divide by 0.09290304 and the answer is 96,875 square feet. Multiply by ten instead and you get 90,000, which is 6,875 square feet short. As a share of the true value that is 6,875 divided by 96,875, or 7.1 per cent.

Go the other way with the same carelessness and the error changes size. A United States show reports 100,000 net square feet. The correct conversion is 100,000 times 0.09290304, which is 9,290 square metres. Dividing by ten gives 9,000, understating by 290 square metres or 3.1 per cent. The asymmetry catches people, because the same mental shortcut produces a 7.1 per cent error in one direction and a 3.1 per cent error in the other, so a portfolio using it inconsistently generates a spread rather than a constant offset.

Rate cards inherit the problem and turn it into money. A stand price of 320 euros per square metre is 320 times 0.09290304, which is 29.73 euros per square foot. An analyst dividing by ten quotes 32.00, overstating the price per square foot by 7.6 per cent. On a 5,000 square foot island that is a difference of 11,350 euros in a comparison an exhibitor is using to decide between two shows.

The half converted calculation is worse again, and I have seen it survive a budget review. Somebody converts the rate to euros per square foot correctly, at 29.73, then multiplies it by an area still held in square metres. A 400 square metre stand comes out at 11,892 euros instead of 128,000, an understatement by a factor of 10.8, and it looks so much like a data error that the person who finds it usually blames the source system and reloads the file. Every one of these failures is a missing unit label, and none of them is a failure of arithmetic.

A benchmark that grows 9.5 per cent because of a rounding rule

Now the case from the opening, with the numbers in it.

The 2025 edition certified 8,400 square metres. The 2026 edition was recorded in the portfolio warehouse in square feet, at 92,000, because that year's figure came in through a United States reporting template. Converted properly, 92,000 square feet is 8,547 square metres, so the show grew by 147 square metres, which is 1.8 per cent.

The analyst converted the 2025 figure to square feet using the ten times shortcut, giving 84,000, and compared it against the 92,000. That produces apparent growth of 8,000 square feet, or 9.5 per cent, and it went into a portfolio review as the second best performing show in the group.

The whole 7.7 point gap is arithmetic. Worse, it is invisible to every sanity check anyone normally runs, because both numbers are in the right units, both are the right order of magnitude, and the growth rate is plausible for a recovering sector. A conversion error that produced 995 per cent growth would have been caught in the first review. One that produces 9.5 per cent gets a slide.

Convert once, store both, label the unit

Three rules, and the third one is the one people skip.

Keep the original value in the unit it was certified in. That is the number an auditor signed, and the moment you overwrite it with a converted figure you have lost the ability to reproduce the certificate.

Store the converted value in a second column, computed once by a function rather than by hand, using the full factor. A spreadsheet with a hardcoded 10.76 in forty different cells will eventually have a 10.7 in one of them.

Record the unit in a column of its own, on every row, with no default. A default is what turns a missing unit into a wrong unit, and area columns are unusually prone to it because a bare 9,000 looks equally sensible in either system. Whether the whole portfolio should standardise on one internal unit is P18's argument and it goes further than a single conversion.

Round only at the point of display, and round the same way in both units. Pick the rounding rule once, write it into the reporting layer, and put it in the same document that defines the figure, so the next analyst inherits a decision instead of making a new one. Publishing 96,875 square feet against 9,000 square metres is consistent. Publishing 96,900 against 9,000 is also consistent. Publishing 90,000 against 9,000 is a different show.

Which direction should you convert in?

For an internal warehouse, hold everything in square metres and convert for display. The reason is the certificate. UFI certifies in square metres, so the metric value is the one with an audit trail behind it, and what belongs inside that certified figure is settled by the standard rather than by your data model.

For a document going to a United States audience, convert to square feet and say so in the same line. A figure labelled "96,875 net square feet (8,999 net square metres, audited)" survives being pasted into somebody else's deck. A bare 96,875 does not.

For a benchmark comparison, convert the smaller side to the unit of the published series rather than the other way round. If you are comparing three of your shows against a United States index built in net square feet, convert your three and leave the index alone, because the index has a published methodology and rounding its components quietly changes what you are comparing against. Choosing an index your show actually belongs to is a separate judgement that P20 works through.

Where this stops

A perfect conversion does not make two figures comparable. It makes them the same unit, which is a smaller claim than it sounds.

The certified 8,547 square metres above and a United States show's 92,000 net square feet are the same quantity of floor and may still be measuring different things, because one may include outdoor plots and unpaid feature space while the other counts only sold indoor booth area. Unit alignment is the first of two steps and the definitional alignment is the harder one.

There is also a precision trap in the other direction. Reporting a converted figure to more digits than the original justifies implies an accuracy nobody has. A certificate reading 8,400 square metres is a rounded number itself, so converting it to 90,417.02 square feet is arithmetically correct and quietly dishonest. Carry the original's precision through: 8,400 becomes about 90,400, and the two zeros are doing honest work.

This week, find every area column in your reporting and check whether a unit is recorded against each row or assumed by whoever wrote the query. Where it is assumed, convert one known figure by hand against the certified value in the published standards and see whether the pipeline agrees with you.