What the dashboard should actually answer
Visitor flow describes the paths people take through a museum or venue: which galleries they enter, the order they visit them, and where they tend to congregate or bypass. Dwell time measures how long someone remains within a defined zone or near a specific exhibit. Together, these two metrics give operations and curatorial teams a picture of how physical space is actually used, rather than how it was intended to be used.
Proximity technology generates these insights through different mechanisms. Bluetooth beacons can passively detect when a device enters and leaves a zone, logging timestamps that the backend converts into dwell-time estimates. NFC taps and QR scans record active interactions, which also carry timestamps but only capture visitors who choose to engage. The two data sources answer different questions: beacon data reveals where people go and for how long, while scan data reveals what they chose to read or view in detail.
For museums, the practical value lies in several areas. Curators can compare dwell times across exhibits to understand which pieces hold attention and which are walked past. Operations managers can identify bottleneck galleries where flow regularly slows, informing signage or layout changes. Event organisers can see whether temporary installations pull visitors away from permanent galleries or simply add to peak congestion. Staffing decisions, such as where to place floor guides during busy periods, can be informed by real flow patterns rather than assumption.
A critical distinction under current UK privacy guidance is between aggregate, anonymised patterns and individual tracking. Visitor flow and dwell-time insights should be derived from data that cannot be linked back to a specific person. This means working with hashed device identifiers or, preferably, counts and durations that never store a raw device address at all. The question is not whether the technology can track individuals, but whether the system is configured so that it does not.
Link technical events to business outcomes
How dwell time is actually measured
With beacons, dwell time is not measured directly. The system detects a device's RSSI value crossing a threshold to register a zone entry, and crossing it again (or a second beacon's threshold) to register an exit. The time between those two events becomes the dwell-time reading. This means the accuracy of the figure depends entirely on how well the thresholds are calibrated for that specific room, its furnishings, and the typical number of people present.
NFC and QR interactions provide a different signal. A tap or scan logs an exact timestamp, and if the same visitor interacts with another tag later, the system can calculate time between interactions. However, this only measures engaged dwell time, not total time in a zone. A visitor may spend eight minutes in a gallery but only scan one QR code in the final minute.
Defining meaningful zones
Zone design is where many projects succeed or fail. A zone that is too small, such as a one-metre radius around a single pedestal, will produce unreliable dwell times because RSSI fluctuation causes constant entry and exit events. A zone that covers an entire wing produces an average dwell time so broad that it obscures the differences between individual exhibits.
A practical approach is to align zones with physical or curatorial boundaries: a single gallery room, a clearly defined alcove, or a cluster of related exhibits. This makes the data easier to interpret and aligns with decisions the museum actually needs to make, such as whether to relocate a exhibit or widen a doorway.
Typical use cases
- Bottleneck identification: Repeated flow data showing that a narrow corridor between two popular galleries consistently slows visitor movement, prompting physical layout changes or one-way routing during peak hours.
- Exhibit performance comparison: Dwell-time averages across similar exhibits revealing which interpretive approaches hold attention longer, informing future curation and label design.
- Layout change validation: Before-and-after flow data after relocating a popular exhibit or adding a new entrance, providing evidence of whether the change had the intended effect.
- Peak-period staffing: Heatmap-style flow data showing which galleries experience the highest density during school holiday periods, allowing floor staff to be positioned where they reduce congestion rather than where they are least visible.
Installation and calibration factors
Beacon placement for flow analysis differs from placement for content triggering. For dwell time, consistency matters more than precision. Mounting beacons on ceilings at a consistent height, away from metal fixtures and air-handling units, produces more stable RSSI readings than wall-mounting at varied heights. Each zone typically requires at least two beacons to distinguish between someone standing near the boundary and someone who has genuinely left the area.
Calibration involves walking the zone with a reference device, recording RSSI values at known distances, and adjusting the entry and exit thresholds until the system's zone transitions match observable movement. Without this step, dwell-time data will contain systematic errors that are difficult to detect later because the numbers look plausible even when they are wrong.
Interpretation, action and review
Treating RSSI-based dwell time as precise
The most common mistake is presenting dwell-time figures to stakeholders as if they were measured with a stopwatch. In reality, a reading of "four minutes and twelve seconds" might mean anything between three and six minutes depending on calibration quality, crowd density, and the visitor's phone model. The useful information is in the comparison between zones and between time periods, not in the absolute seconds. When reporting, use rounded averages and ranges rather than false precision.
Confusing interaction rate with dwell time
A high QR scan rate does not mean a high dwell time, and vice versa. An exhibit might attract long gazes but few scans because the label is unclear or the QR code is poorly placed. Another might generate many taps on an NFC tag from visitors who stop briefly, interact, and move on. These are different metrics and conflating them leads to incorrect conclusions about what is working.
Ignoring the invisible majority
Beacon-based flow data only captures visitors who have Bluetooth enabled and, in most current deployments, who have the venue's app installed. Web-based Bluetooth approaches exist but have significant browser and operating-system limitations. This means the data represents a subset of visitors, and that subset may skew towards younger demographics or repeat visitors. It is misleading to present flow data as a complete picture of all footfall without stating the detection rate and its likely bias.
Dwell time without context
A long dwell time is not always positive. A visitor may be lingering because they are deeply engaged, because they are lost, because they are resting, or because a crowd ahead of them is blocking the path. Without supplementary observation, such as periodic manual counts or visitor feedback, the data alone cannot explain the why behind the number.
Key checks for suppliers and systems
- How is a dwell event defined technically? Ask for the exact logic: single-beacon threshold, multi-beacon consensus, minimum duration before a dwell is logged, and what happens during brief signal drops.
- What happens at zone boundaries? Some systems produce phantom entries and exits when a visitor stands on the edge of a zone. Ask how boundary behaviour is handled and whether there is a hysteresis setting to prevent flicker.
- Can you export anonymised aggregate data? You need zone-level counts and average durations, not individual device logs. Confirm that exports contain no raw MAC addresses or advertising identifiers.
- How does the system handle signal noise from crowds? A room with fifty people behaves differently from an empty room. Ask whether the system adapts or whether calibration is only valid for a specific density range.
- What is the minimum viable pilot? A single-gallery pilot with manual observation alongside sensor data is the most reliable way to validate that the system's outputs match reality before committing to a wider rollout.
The practical next step is to select one gallery where the museum already has questions about visitor behaviour, install a small number of beacons with careful calibration, and run the system alongside a few hours of manual observation. Comparing the sensor data with what staff can see on the ground reveals whether the system is producing trustworthy enough insights to justify expanding to other spaces.
