Metrics that match the operational outcome
Footfall and dwell-time metrics are the two most fundamental measurements in physical location analytics. Footfall is a count of devices detected within a defined zone over a specific period. Dwell time measures the duration a detected device remains within that zone. Together, they form the baseline for understanding how people move through and interact with a physical space, whether that is a retail floor, a museum gallery or an event hall.
Proximity technologies—primarily Bluetooth Low Energy (BLE) beacons, but also Wi-Fi access points—derive these metrics by listening for signals from nearby smartphones. When a device enters a beacon’s range, the system records a "first seen" timestamp. When the signal disappears or drops below a defined threshold, it records a "last seen" timestamp. The difference between these two points provides the raw dwell time.
However, the raw data requires interpretation. A system does not count people; it counts devices. A single visitor carrying two phones registers as two footfall entries. A group of three people walking together, only one of whom has Bluetooth enabled, registers as one entry. Understanding this distinction is the first practical step in evaluating any location analytics platform.
The calculation of dwell time also depends heavily on how the system handles signal gaps. Bluetooth signals fluctuate constantly due to interference, human bodies absorbing the signal and the device being moved in a pocket or bag. If a beacon misses three consecutive pings from a phone, the system must decide whether the person has left the zone or simply experienced a momentary signal drop. Most platforms use a "gap tolerance" or "timeout" setting—for example, ignoring gaps of under 30 seconds before declaring the device has left. The strictness of this setting directly alters the reported dwell time.
Collect comparable and interpretable evidence
Retail environments
In retail, footfall answers basic capacity questions: how many people entered the store, and at what times? Dwell time adds behavioural context. If visitors to a specific promotional display have an average dwell time of five seconds, the display is likely failing to capture attention. If the dwell time averages two minutes, the content is engaging, but the operational team must then check whether that engagement translates to sales or merely creates a bottleneck. Comparing dwell time against point-of-sale data reveals the conversion efficiency of specific zones.
Museums and galleries
Museums use dwell time to evaluate exhibit engagement without requiring visitors to interact with a screen or scan a code. By placing beacons near key exhibits, curators can measure average viewing times. If an exhibit intended to hold attention for three minutes shows a median dwell time of twenty seconds, it suggests the labelling, positioning or content needs review. This data is particularly useful when planning temporary exhibitions, as it provides objective evidence of where visitors spent their time versus where they rushed through.
Events and venues
For event organisers, footfall and dwell time help identify congestion and underutilisation. High footfall with very low dwell time in a corridor indicates a flow problem. High footfall with high dwell time in a catering area might indicate long queues rather than high engagement. During multi-day events, tracking dwell time across different sessions helps organisers understand which speakers or topics held the audience's attention longest, informing future programming.
Technical configuration
The accuracy of these metrics depends on the beacon’s advertising interval. A beacon set to broadcast every 100 milliseconds generates highly granular data but drains the battery rapidly. A beacon broadcasting every 1000 milliseconds preserves battery life but introduces coarser time-stamping, making short dwell times harder to measure accurately. Operational managers must balance the required granularity against the maintenance burden of more frequent battery replacements. For a museum exhibit where visitors typically stay for minutes, a slower interval is usually sufficient. For a fast-moving retail entrance, a faster interval captures brief passes more reliably.
Verify that the measure remains useful
The one-device-one-person assumption
The most persistent mistake is treating device counts as people counts. In environments with high staff presence, staff devices will inflate footfall unless filtered out. In family-oriented venues, a group of four might register as one or two devices. When evaluating a system, ask the supplier how they handle device-to-person estimation and whether they offer manual calibration factors to adjust raw counts towards observed reality.
The ping-pong effect
When a visitor stands near the boundary of a zone, the fluctuating RSSI can cause the system to rapidly log them as entering and leaving. This "ping-pong" effect fragments a single visit into multiple short visits, destroying the dwell-time data. Check whether the analytics platform applies hysteresis or smoothing algorithms to prevent a device from being logged out and back in within a short window.
MAC randomisation
Modern iOS and Android operating systems use MAC randomisation, changing the Bluetooth identifier periodically to prevent long-term tracking. This limits the usefulness of footfall and dwell-time metrics for return-visit analysis. A device that visits on Monday and Wednesday will likely appear as two separate, anonymous entries. These metrics are best used for aggregate, anonymous behaviour within a single session or day, not for building individual customer profiles.
Verification during a pilot
Before trusting the numbers at scale, verify them manually. During a pilot phase, position a team member at a zone entrance with a manual clicker for a set period. Compare the manual count against the system’s footfall report. Then, select a few individuals, note the exact time they enter and leave a zone, and compare your stopwatch reading against the system’s dwell-time log. If the system reports a 90-second dwell time but your stopwatch shows 120 seconds, the gap tolerance or advertising interval needs adjustment.
Key questions for a supplier
- How does the platform define a zone entry and exit—by a single RSSI threshold or a rolling average?
- What is the default gap tolerance, and can it be adjusted per zone?
- How does the system filter out staff devices or stationary devices left in the space?
- Can the platform export raw timestamp data, or only pre-aggregated averages?
- How are metrics affected by devices that have Bluetooth switched on but location services disabled?
Footfall and dwell-time metrics provide valuable operational insight, but only when the underlying data collection mechanics are understood and calibrated for the specific physical environment. Treating the dashboard figures as precise truths without testing them against observed reality is a reliable way to make poor operational decisions.
