Start with a decision and a baseline
A proximity pilot is fundamentally a learning exercise, yet many projects fail because their key performance indicators (KPIs) are borrowed from mature digital marketing programmes rather than shaped for a physical-space trial. The purpose of defining KPIs at the outset is to establish what constitutes a useful result, not to guarantee a return on investment before the hardware is even mounted.
There is a critical distinction between a metric and a KPI in this context. A metric is a raw data point, such as the number of times a beacon’s signal was detected by a passing smartphone. A KPI ties that data point to a specific business or operational question: for example, whether zone-based detection correlates with a measurable change in visitor dwell time. Defining KPIs means selecting the few metrics that actually answer the pilot’s core questions and setting realistic thresholds for them.
Because a pilot runs in a single, finite physical environment, its KPIs should prioritise learning over scale. If the objective is to test whether Bluetooth beacons can reliably trigger indoor navigation prompts in a crowded exhibition hall, the primary KPI should focus on trigger consistency and positional accuracy within that specific venue, not on total user volume. Volume-based KPIs become relevant later; during a pilot, they often mask technical or environmental problems.
Compare periods, zones and audiences fairly
The right KPIs depend entirely on what the pilot is trying to prove. Different operational contexts demand different definitions of success, and the infrastructure required to measure them varies significantly.
Retail environments
In a retail pilot, the business question is usually whether proximity-triggered content influences behaviour at or near the point of sale. A well-defined KPI might be the redemption rate of a specific offer triggered in a designated aisle, compared against a baseline measured before the beacons were active. Another valid KPI is the change in average dwell time within a defined zone, provided the venue has a reliable method of establishing that baseline (such as existing footfall counters) rather than relying solely on beacon detection, which only measures opted-in devices.
Museums and cultural venues
For museums, pilots often focus on content delivery and accessibility. A suitable KPI here might be the percentage of visitors who complete a recommended route, measured by the sequence of zones their device registers. Alternatively, if the pilot is testing an audio guide trigger system, the KPI could be the ratio of successful content triggers to expected triggers based on known footfall patterns. The emphasis is on content reach and route adherence rather than direct revenue.
Events and temporary venues
Event pilots face unique constraints: the infrastructure is temporary, crowd density fluctuates wildly, and there is no opportunity for a prolonged baseline period. A practical KPI for an event might be the average time taken for a user’s device to receive a wayfinding prompt after entering a designated zone during peak hours. This tests the system’s performance under stress. Another is the percentage of session check-ins accurately logged by the proximity system versus manual counts at the door.
Technical versus business KPIs
Every proximity pilot needs both categories, but they should not be weighted equally when evaluating success. Technical KPIs—such as beacon uptime, battery drain over the pilot period, or signal consistency in a specific RF environment—validate the infrastructure. Business KPIs validate the use case. A pilot can succeed technically (the beacons broadcast reliably) but fail commercially (the triggered notifications had no measurable effect on behaviour). Both outcomes are valuable, provided the KPIs were defined clearly enough to separate the two.
Record conclusions and next tests
The most frequent error in defining pilot KPIs is setting targets based on manufacturer specifications rather than measured environmental conditions. A beacon data sheet may state a range of 70 metres in open air, but in a Victorian museum with dense wall materials and high footfall, the effective reliable range may be far shorter. If a KPI assumes open-air performance, it will be meaningless. KPIs must be grounded in the physical reality of the venue, which usually requires a preliminary RF survey before the pilot KPIs are finalised.
Planning note: Use figures as starting assumptions only, then replace them with measurements from the actual mounting position and representative devices.
Another common mistake is ignoring the consent filter. Proximity technology in the UK operates within a framework that requires user consent for location tracking and personalised notifications. If only a small fraction of visitors opt in, aggregate metrics like total zone entries will look low, not because the technology failed, but because the consent strategy did. A robust set of pilot KPIs should include a metric for the opt-in rate itself, treating consent as a variable to be tested rather than a given.
Confusing correlation with causation also undermines pilot evaluations. If dwell time increases in a beacon-monitored zone, the KPI framework must account for other variables: seasonal traffic, adjacent promotions, or physical layout changes. Without a control zone or a pre-pilot baseline measured by independent means, the KPI cannot reliably attribute the change to the proximity system.
Key checks before finalising KPIs
- Baseline existence: Can the target metric be measured without the proximity system active? If not, you cannot calculate the change your pilot caused.
- Consent alignment: Does the KPI rely on data that requires explicit opt-in, and has a realistic opt-in rate been factored into the target?
- Environmental constraints: Has an RF survey been conducted to ensure the KPI targets are physically achievable in the specific venue?
- Separation of concerns: Are technical KPIs (beacon health, signal stability) clearly separated from business KPIs (behaviour change, content engagement)?
- Data minimisation: Does the KPI require collecting more personal data than is necessary to answer the pilot’s question? Under UK guidance, the data collected should be the minimum required for the stated purpose.
Defining KPIs for a proximity pilot is an exercise in disciplined scepticism. The goal is to set targets that prove whether the technology works in your specific physical space, for your specific audience, under your specific consent model. If the KPIs are vague, optimistic, or detached from the venue’s RF environment, the pilot will produce data that cannot support a confident decision to scale or abandon the project.
