Measuring museums: impact, not admissions - what an exhibition actually achieved
A ticket count tells you how many people came through the door. About an exhibition, it tells you almost nothing.
After every temporary exhibition, the board, the funding body and the sponsors ask three questions. Did it bring more people? Did it bring different people? Or the same people, who stayed longer? The admissions figure answers the first at best - and only if there are no free days, no annual passes and no second entrance. Plenty of museums have all three.
That makes museums a special case. Retail has a till, transport has a timetable system, offices have a booking tool. In a museum, the sensor on the ceiling is often the only data source that knows who was where and when. This article goes through what a museum can make of that. It builds on Measuring dwell time, which explains what a zone is, and on 800 visitors is not an answer, which is about never reading a figure without its surroundings.
Three questions for an exhibition
| Question | Metric | What matters |
|---|---|---|
| More people? | Visitors during the exhibition against a baseline period | periods of equal length, the same weekdays, a significance block rather than a hunch |
| Different people? | Age and gender distribution per period | a distribution across all observations, not a share of visitors |
| Stayed longer? | Dwell time and engagement per room | every room a zone, measured by the sensor rather than estimated |
The first row is a period comparison. How to read a significance block is covered in The visitors who don't buy. In a museum it matters even more than in a shop, because the numbers are smaller. An eight per cent rise in a museum with 300 visitors a day may be chance, and a report to the board of trustees should say so before somebody else does.
The second row calls for careful wording. “The exhibition brought in a younger audience” is a statement about a distribution, not about heads. Why every demographic figure needs a denominator is explained in What a counter can tell apart.
The third row is the one a ticket count is blind to. A museum can have exactly last year's attendance and still be showing a better exhibition - if people spend four minutes longer in the main gallery on average.
Which room holds people
Dwell time per room is the closest thing to measuring impact. It is measured where it happens: each gallery becomes a zone, and the sensor reports when someone enters and leaves it. If a single masterpiece is to be assessed, the area in front of it gets a zone of its own. The work then has a figure of its own alongside the room.
Dwell time reads best alongside how many people enter a room at all. A room with few but long visits is a different thing from a passage room where everyone stays briefly. The four behaviour stages and the engagement rate that tell them apart are in Measuring dwell time.
For programme planning, this is the most valuable figure in the building. Rooms that hold visitors become the argument for the next exhibition. Rooms everyone merely walks through show where the hang, the labelling or the route deserves a second thought.
A visitor figure that stands up to the funding body
The funding application, the annual report, the press release: all of them need a figure that holds when someone asks how it was arrived at. In historic buildings that is harder than elsewhere. Several entrances, high ceilings, narrow passages, heritage protection. Turnstiles are out of the question, and counting by hand is an estimate.
A ceiling mount solves most of it. The 3D sensors we use are available for mounting heights from 1.95 to 20 metres; according to the manufacturer, they count with up to 99 per cent accuracy even in dense situations. The sensors on every entrance are combined into a single visitor figure for the building, with no construction work. What other counting methods exist, and why their accuracy claims can't be compared, is covered in Which technology counts people how accurately?.
What makes the figure robust is what surrounds it: count series over two years, export at any resolution, and a correction that is documented with a reason and an approval rather than overwriting a cell in a spreadsheet. Anyone running several museums compares them on the same basis - how to do that fairly is in Comparing sites.
Who in the building needs what
In a museum, the same data answers questions from five different desks. Most are described in detail in earlier articles, so here is the overview:
| Desk | Question | Answer |
|---|---|---|
| Gallery attendants | Can the next group go into the gallery? | occupancy per room against capacity, live, showing “stale” rather than “empty” |
| Front desk | When does the ticket desk need a second person? | estimated wait against a threshold, an alert before the ten-minute mark |
| Operations | How many attendants at 2 pm on a Sunday? | recommended staffing per hour from the measured footfall |
| Building services | Why are the visitor lights on in an empty gallery? | lighting and ventilation follow occupancy per room |
| Communications | When should we invite people to come? | the quietest hours, ready to pass on to visitors |
The groundwork: occupancy and its drift in Occupancy is a balance, not a count, waiting time and thresholds in Managing queues, the staffing recommendation from the 75th percentile in The visitors who don't buy, and the link to building services in Occupied versus paid-for space.
One point deserves particular attention in a museum. Conservation-grade climate control runs around the clock, and rightly so. But visitor lighting and visitor ventilation need not run when nobody has been in the room for an hour. A call to the building management system when a room becomes empty or occupied again is enough. The only thing that matters is that “no data” is passed on as unknown rather than as “empty” - otherwise the lights go out because a sensor has gone quiet.
Admission with a traffic light
An animal house, a reading room, a small cabinet exhibition: many spaces have a maximum occupancy, and whoever stands at the door sees the queue, not the people inside. A traffic light at the entrance solves this: green means go in, red means wait.
All it takes is a public display at the entrance or at the start of the tour, with no sign-in, showing a traffic light and a breakdown per room. Attendants and visitors see the same figure, and the empty gallery next door becomes a recommendation rather than a missed opportunity.
Related venues
What applies to a museum applies, with variations, to three related kinds of place:
- Cinema and auditorium. Tickets sold and people in the room are two figures. Reconciling them per screening - actual against expected, deviation in per cent, 13 months of history - shows where complimentary tickets, double bookings or gaps in ticket checks are. The method is described in Occupancy is a balance, not a count.
- Theme park. The wait-time board is the most-read sign in the park. What a measured rather than guessed board can achieve is shown by Freizeitpark Plohn, described in Managing queues. Throughput per attraction comes on top, showing where a ride's cycle is getting longer.
- Outdoor landmarks. The square, the bridge, the viewpoint: according to the manufacturer, weatherproof outdoor models also count in the open, from 9 lux and between −33 and +40 °C. The city and the tourism organisation get a figure that is the same for everyone, and a visitor-management measure gets a before and after with significance.
Where a museum should start
- The entrances. One visitor figure for the building, combining every door. It is the basis for everything else and replaces the estimate in the annual report.
- The rooms decisions depend on. The temporary exhibition gallery plus two or three collection galleries for comparison. A first exhibition assessment needs no more zones than that.
- A baseline period. Measure for at least a few weeks before the opening. Without a baseline there is no comparison, just a number.
- Then operations. Ticket desk, attendants, building services, once the first weeks show where the peaks are.
What belongs in the tender
- Are the sensors on every entrance combined into a single visitor figure for the building?
- Can every gallery, and the area in front of a key work, be set up as a zone of its own?
- Is there a period comparison with a significance statement, and an export for the report to funders and sponsors?
- Are demographics shown as a distribution per period, with a denominator?
- Does occupancy per room show a room without data as stale rather than empty?
- Is there a public display without sign-in, and can access to it be revoked?
- Can building services retrieve occupancy per room, or be notified when it changes?
How we do it
The ANALYSIT Counting System adds up every sensor on a site into a single visitor figure, shows count series at any resolution with peak hour, year-on-year comparison and a portfolio benchmark across several venues, and exports them as CSV and XLSX.
For assessing an exhibition, ACS compares a baseline period with the exhibition period - visitors and groups with daily values, target rings and a significance block, exported as CSV, XLSX and PDF. Dwell time and engagement are shown per room, the age distribution across seven classes and the gender distribution per period.
In operations, ACS shows occupancy per room against capacity with a five-minute trend and a refresh every 10 seconds, including as a public display without login. Waiting time at the ticket desk and cloakroom runs against a target and a critical threshold, with alerts over five channels; the staffing recommendation per weekday and hour comes from the 75th percentile of footfall; and display triggers call the building management system by webhook when a room becomes occupied or empty. For auditoriums and cinemas, capacity verification reconciles actual against expected per screening, with 13 months of history.
If your next exhibition is coming up and you want to measure it this time, talk to us. Ideally before it opens - because of the baseline.