Exposure Analytics has published the first behavioural benchmark for physical spaces. The framework is the important part, and the dataset underneath it needs reading carefully before anyone starts quoting the headline number.
TL;DR The Physical Engagement Index 2026 benchmarks 43.8 million measured visitors across more than 3,000 events since 2020. It reports a global engagement rate of 35.59%, average dwell of 13 minutes 50 seconds among engaged visitors, and a 20 times gap between the strongest and weakest spaces. The framework matters more than the headline: volume, engagement and dwell are three separate questions, and the industry has only ever answered the first.
Every exhibition conversation I have sat in over the past 20 years arrives at the same number. How many people came. It is the number shows sell on, exhibitors buy on and boards approve on, and it tells you the least about whether any of it worked.
Exposure Analytics has just published something that tries to fix that. The Physical Engagement Index 2026 is the first annual behavioural benchmark for physical spaces, built on sensor data from more than 3,000 events and 43.8 million visitors between 2020 and April 2026, across 18 countries and 18 sectors.
I should be straight about the relationship. Rob Murdoch, who founded Exposure Analytics, is someone I know, and he asked me to look at a draft of this report editorially before publication. I have no commercial interest in the business and nobody has paid for this piece. Exposure Analytics answered a set of questions from me for this article, including two that produced figures not published in the report itself. What follows is what I think the Index gets right, what it does not yet prove, and what happens to the events industry if the argument lands.
What is the Physical Engagement Index?
The Physical Engagement Index is an annual benchmark published by Exposure Analytics that measures behaviour inside exhibition halls, retail floors, activations and brand spaces. It reports three signals: how many people were present, how many stopped, and how long the ones who stopped stayed. The 2026 edition is the first.
The report’s own framing is blunt. Footfall tells you the room was full. It does not tell you what happened in the room.
That is not a new complaint. Anyone who has stood on a stand at four in the afternoon watching several thousand badge holders walk past without breaking stride already knows footfall is a presence metric wearing a performance metric’s clothes. What is new is a dataset large enough to say so with numbers.
The report cites Splash research showing 67% of marketers still rate attendance among the most important metrics to their business. That is the gap the Index is aimed at.
What does the framework actually measure?
The framework splits physical performance into three questions. Volume is how many were present. Engagement is how many stopped rather than passed through. Dwell is how long the engaged visitors stayed. Exposure Analytics positions these as the measurement layer sitting between presence and commercial outcome.
Three signals rather than one is the whole idea, and it is a good one. Volume, engagement and dwell are not different views of the same thing. They are different decisions. A space with high volume and low engagement has a threshold problem. A space with high engagement and low dwell has a content problem. A space with low volume and strong performance on both is probably in the wrong hall.
Two definitions matter before anyone quotes the numbers.
Dwell is measured among engaged visitors only. The headline 13 minutes 50 seconds isn’t what the average visitor spends in a space. It’s what the average visitor who stopped spends. The methodology is clear on this. The headline is easy to misread, and it will be misread.
Engagement is a sensor-defined behaviour, not a state of mind. It means a visitor interacted with the space rather than passing through it. That’s measurable, and it’s a great deal more than footfall gives you. It is not the same thing as interest, intent or qualification, and the space between the two is where a lot of bad sales collateral is going to get written over the next 18 months.
How engagement is defined, and why that matters
The report states the mechanism. Engagement is set by distance from the sensor and by a specified length of time, and both are set by the client for each deployment.
Publishing that is to the company’s credit. Most measurement vendors do not tell you where the line sits. It also carries a consequence the report doesn’t draw out. If the zone boundary and the time threshold move from client to client, engagement rate is not a like-for-like measure across every deployment in the way a turnstile count would be. A 12 square metre stand measured at a few seconds inside two metres and a flagship retail floor measured at half a minute inside ten are both producing something called an engagement rate, and they are not the same number.
That is not a reason to discount the benchmark, and client-set zones are operationally sensible. A small stand and a department store are not the same measurement problem. But if the Index is going to work as an industry benchmark rather than a client reporting tool, the second edition needs to publish the range of thresholds in use, or set a standard one and report against it. The first edition does not publish that range. The rest of the framework is sound enough to deserve that it does.
Read the sample before you read the benchmark
Of 43.8 million measured visitors, 33.2 million came through UK and Ireland sensors and 33 million were measured outdoors. Exhibitions account for 9.6 million. The global average of 35.59% is therefore not an exhibition benchmark and should not be used as one.
This is the part I would want any event professional to read twice, and to the report’s credit it publishes the numbers that let you do it.
Three-quarters of the dataset is UK and Ireland. The global average is, in practice, a UK average with international texture around it. That is not a flaw, it reflects where the sensors are, and the report says as much. It does mean the useful comparison for a British organiser is the UK line, and for a German exhibitor the German one, which at 48.9% is the highest engagement rate of any country with more than a million measured visitors.
By setting, the split is 33 million outdoors against 10.7 million indoors. The Index defines four types of physical space and publishes benchmarks for three of them. Exhibitions account for 9.6 million measured visitors, activations 26.6 million and retail 130,000. The remaining 7.5 million sit under events, the broader occasion category covering festivals, conferences, launches and hospitality, which the report defines but does not give its own line in the interaction table.
So when the exhibitions industry picks this report up, the number to quote is not 35.59%. It is 39.7% engagement and 16 minutes 54 seconds dwell, which is what exhibitions specifically returned. Both sit above the global average, which is worth knowing on its own.
The retail figure of 54.8% comes off 130,000 visitors. That is directionally interesting and nothing more, and the report is careful to say so about its smaller cuts.
Why the sector table is mostly an activation table
Home tech and automotive are the two largest sectors in the Index, together accounting for roughly 73% of all measured visitors. Exposure Analytics confirmed to Event Tech Live that home tech data is 98% activations and 1.3% exhibitions, and that automotive is 72% activations against 16% exhibitions.
This is the single most important thing I would tell anyone picking up the sector rankings, and it is not in the report.
Home tech carries 18.4 million measured visitors and automotive 13.7 million. Between them that is 32.1 million of the 43.8 million total. When I asked how those sectors split by space type, the answer was that home tech is almost entirely activation data, at 98%, with exhibitions accounting for 1.3%. Automotive is 72% activations and 16% exhibitions.
So the two sectors carrying most of the Index are, in practice, brand activation numbers. Home tech’s six minutes 51 seconds dwell, the shortest of any major sector, is not a verdict on how home tech brands perform at trade shows. It is a verdict on how they perform in activations, which are built for a different job and return roughly half the dwell of an exhibition across the whole dataset.
That is not a criticism of the data. It is an instruction for reading it. Take the interaction type table first, work out what kind of space you actually operate, then read the sector line through it. A pharmaceutical team looking at 15 minutes 50 seconds of dwell and an automotive team looking at 11 minutes 46 are not looking at comparable environments, and the sector table on its own does not tell you that.
It also sharpens the report’s own best argument. If the biggest sector numbers are activation numbers, then the exhibition’s figure of 39.7% engagement and 16 minutes 54 seconds dwell is doing more work than its 9.6 million visitors suggest. Exhibitions are outperforming the headline, and the headline is being pulled down by a format that was never designed to hold anyone for a quarter of an hour.
Why the 20 times gap is the actual finding
The top 10% of measured spaces converted 76.3% of visitors into engagement. The bottom 10% converted 3.9%. On dwell, the top decile held visitors for 33 minutes 31 seconds against three minutes 52 seconds at the bottom. Footfall measurement cannot see either difference.
Everything else in the Index is context for this. A 20 times spread between best and worst inside a single category is not normal. Digital advertising does not report ranges like that within a channel. Retail conversion does not either. A spread that wide is usually the signature of a category where the measurement has been crude enough to flatten real differences, which is precisely the argument the report is making.
Two qualifiers sit on the number. The first is the threshold question. With engagement zones and time settings specified per client, some of the distance between deciles could be specification rather than behaviour. The cohorts are large, 2.2 million visitors in the top decile and 2.4 million in the bottom, which spans a lot of deployments and makes a pure measurement artefact unlikely. It cannot be ruled out on what is currently published.
The second is the explanation. The report attributes the difference to intent rather than budget. Strong spaces are built around one clear behavioural objective, engage more than one sense, lower the friction of walking in, and give people a reason to stay. Weak spaces are visually busy and behaviourally vague, full of objects and empty of purpose.
I think that’s probably right. I also think it’s the least evidenced claim in the document. Sensors measure what people did, not why a space was designed the way it was. The gap is measured. The explanation is inferred. Pairing the top and bottom cohorts against stand size, position and sector would turn a persuasive argument into a proven one.
Exposure Analytics has told Event Tech Live it is now running that analysis. It is the number I will be waiting for.
Why behavioural measurement matters now
The data event organisers rely on most, registration data, is getting less reliable at exactly the moment behavioural measurement is becoming affordable. Registration records stated intent at a single point in time. Sensors record what actually happened on the day.
Registration data has always been soft. People give the answer that gets them through the form. People register three months out and arrive with a completely different reason for being there. Job titles change. Priorities change. Companies restructure.
That data is now being used for matchmaking, personalisation and pre-show recommendations, which means the softness compounds rather than sitting harmlessly in a database.
And there is a newer problem. Attendees are starting to hand registration to AI agents. I have seen it happen. Someone asks their assistant to get them registered, and the form is completed from whatever the agent already knows rather than from what the organiser is trying to learn. That behaviour is not going to slow down.
Behavioral measurement sidesteps all of it. It doesn’t ask anyone anything. It records what happened.
That is a bigger argument than the Index makes for itself, and it is the reason I think this framework matters beyond the company that published it.
What changes if behaviour becomes the metric
Four things move: how floor space is priced, how stands are briefed, when budget is spent, and what organisers are able to sell. Each one is currently anchored to a metric that only measures presence.
Floor plan pricing
Front-and-centre positions carry a premium on the assumption that traffic converts. The Index reports front-left at 49.9% engagement against front-centre at 41.6%, with front-left also holding visitors longer. The caveat is real. Front-centre carries 12.2 million measured visitors against 109,000 front-left, so nobody should rewrite a floor plan on that base alone. But the direction matches what behavioural science would predict, and Patrick Fagan’s contribution in the report makes the mechanism properly: the busiest aisle is the worst place to ask somebody to concentrate. If it holds at scale, the premium the industry charges for prime position is being charged for the wrong thing.
Square metre pricing
Joost Bijsterveld’s piece in the Index argues that the exhibition trade sells by the square metre because that is how it has always sold, and that brand teams defend big footprints because size is the easiest thing to defend to a board. His conclusion, that intent beats size and that size only multiplies what is already working, is the sharpest paragraph in the report. Coming from someone who ran global experiential marketing at Philips, it carries.
When budget gets spent
The Index reports August as the strongest month for engagement at 47.3%, with October second at 44.0%. December sits at 31.4%. That is a spread of nearly 13 percentage points between October and December, with October holding engaged visitors for 16 minutes 10 seconds against December’s 12 minutes 43. The same activation, moved by two months, performs materially differently on both measures.
Almost no event budget I have seen is signed off with reference to seasonal engagement performance. Availability, cost and audience scale, yes. Behaviour, no. For B2B sectors with long planning cycles the report reads October as the strategic peak, which will surprise nobody who sells exhibition space, but now there is a number attached to the instinct.
What organisers can sell
This is the one organisers should care about most. An organiser who can hand an exhibitor behavioural data has a fundamentally different renewal conversation than one who can hand over an attendance figure. Most rebooking discussions currently run on visitor numbers and a hope that the exhibitor’s own lead count came out well. Shared behavioural data turns that into a joint problem rather than an argument across a table.
What the first edition leaves open
Three questions carry into the next edition: whether the 2025 engagement dip is a trend or a sample effect, whether long-dwell sectors are being judged on the wrong primary metric, and whether engagement thresholds should be standardised across deployments.
Engagement fell to 30.3% in 2025 from 43.4% in 2024, on near record measured volume. The report reads that as fractured attention arrives from outside the room, and I have some sympathy with the argument. It is also the kind of movement a change in client or format mix could produce on its own, particularly in a dataset where two activation-heavy sectors carry most of the weight. One year is not a trend.
Culture is the most interesting anomaly in the document. It sits at 22.92% engagement, near the bottom of the sector table, and 39 minutes six seconds dwell, the longest of any sector in the report by a distance. Read on engagement it looks like an underperformer. Read on dwell it is the strongest performer in the dataset. The Index raises the question of which metric should be primary for browsing sectors and stops short of answering it. It is the right question, it deserves a proper answer next year, and plenty of exhibition spaces have exactly the same shape.
The threshold question is the third, and it is the one that decides whether the Index becomes a benchmark the industry can argue with. A published standard, or a published range, would settle it.
What to do with the Index now
For exhibitors, stop accepting visitor numbers as evidence and start asking what behavioural data is available for the hall you are buying into. Then brief your contractor in visitor outcomes rather than square metres. What should a visitor do, feel or understand, and how will you know whether they did. And read the exhibition line, at 39.7% and 16 minutes 54 seconds, rather than the global one.
For organisers, measure a hall rather than a door. Exhibitors are already buying measurement independently at the larger shows. Organisers who provide it become the source of truth. Organisers who do not become the thing exhibitors measure around.
For the measurement vendors, and there is more than one, the Index does something useful for all of them. It establishes that this is a category with a benchmark rather than a set of competing point products. That is worth more to the field than the specific numbers are.
The first edition makes the argument. Whether the industry acts on it is the thing the second edition will actually measure.
Frequently asked questions
What is the Physical Engagement Index?
An annual benchmark published by Exposure Analytics measuring behaviour inside physical spaces. The 2026 first edition covers more than 3,000 events and 43.8 million visitors between 2020 and April 2026, across 18 countries and 18 sectors.
What is a good engagement rate for an exhibition stand?
The Index reports 39.7% average engagement for exhibitions specifically, against a global average of 35.59% across all space types. The top 10% of measured spaces reach 76.3%.
How is engagement measured?
By sensor. A visitor counts as engaged when they remain within a set distance of the sensor for a set length of time. Both the distance and the time threshold are specified per deployment by the client, so figures are most comparable within a given measurement specification.
How is engagement different from footfall?
Footfall counts everyone present. Engagement counts only those who stopped and interacted with the space rather than passing through. In the Index dataset, 15.6 million of 43.8 million measured visitors engaged.
How long do visitors spend on an exhibition stand?
The Index reports average dwell time of 13 minutes 50 seconds, measured among engaged visitors only rather than all visitors. Exhibitions specifically average 16 minutes 54 seconds.
Is the Physical Engagement Index a global benchmark?
It covers 18 countries, but 33.2 million of the 43.8 million measured visitors were recorded in the UK and Ireland. Regional and country figures should be read against the visitor volumes published alongside them.
Where can I download the Physical Engagement Index 2026?
It is available free at physicalengagementindex.com.
Disclosure
Adam Parry reviewed a pre-publication draft of the Physical Engagement Index 2026 at the request of Exposure Analytics founder Rob Murdoch. Neither Event Tech Live nor Adam Parry holds a commercial relationship with Exposure Analytics. This article was not paid for, and was not submitted to the company for approval before publication. Exposure Analytics answered factual questions from Event Tech Live during its preparation.
