Distinguish technical faults from design faults
Most accuracy problems in beacon projects start before a single device is mounted. A manufacturer's data sheet might state a typical accuracy range under controlled conditions, and that figure gets repeated in a business case, a board presentation, or a supplier proposal. By the time it reaches the operational team responsible for delivery, the number has become a commitment rather than a starting point.
The root cause is usually a misunderstanding of what RSSI (Received Signal Strength Indicator) actually measures. RSSI is a measure of signal power at the receiver, expressed in negative decibels relative to a milliwatt. It is not a direct distance reading. A phone or gateway interprets RSSI alongside a propagation model to estimate how far away the beacon might be. That model assumes free-space conditions: no walls, no people, no metal racking, no reflective glass. Few real venues resemble an anechoic chamber.
When a stakeholder asks "how accurate will this be?", the honest answer depends on the specific room, the beacon model, the transmit power setting, the advertising interval, the receiving device, and what else is broadcasting in the 2.4 GHz band at that moment. Until those variables are measured on site, any single-figure accuracy claim is speculative.
The practical consequence is not merely disappointment. Over-promised accuracy leads to incorrect zone definitions, poorly calibrated triggers, and dashboards that show visitors in places they clearly are not. When stakeholders lose trust in the location data, the entire project's credibility suffers, even if the underlying technology is working as well as it reasonably can.
Test likely causes in a sensible order
The key distinction is between positional accuracy and functional accuracy. Positional accuracy asks "where exactly is this person?" Functional accuracy asks "can the system trigger the right action at the right moment?" For most proximity deployments, functional accuracy is what matters, and it is far more achievable.
Retail environments
A retailer might want to trigger a notification when a customer enters the footwear aisle. The stakeholder does not need to know the customer's position to within thirty centimetres. They need to know, reliably, that the customer has moved from the general menswear area into the footwear zone. If the beacon placement and calibration ensure that the trigger fires when a phone is genuinely in that aisle and does not fire when it is two aisles away, the system is functionally accurate regardless of what the estimated distance readout says.
The mistake is translating "we need to know which aisle" into "we need one-metre accuracy." Those are different requirements. Framing the project around the actual trigger logic, rather than an abstract precision figure, sets expectations that can be met and verified.
Museums and galleries
Museum projects often stumble when stakeholders expect exhibit-level detection from a ceiling-mounted beacon. If two exhibits are one and a half metres apart, and the environment includes display cases, metal fixings, and fluctuating visitor density, a single beacon cannot reliably distinguish between them. The practical response is either to zone at room level and use NFC or QR for exhibit-specific content, or to increase beacon density and accept the calibration and maintenance burden that comes with it.
Stakeholders need to see the floor plan with realistic zone boundaries drawn on it, not a marketing diagram showing pinpoint dots over each exhibit. A zone map that shows overlapping detection areas and uncertain boundaries is more useful than a clean diagram that implies precision the system cannot deliver.
Events and temporary venues
Event organisers sometimes request seat-level or stand-level accuracy in exhibition halls. These spaces are among the most challenging for Bluetooth: vast open areas with high ceilings, hundreds of interfering devices, temporary structures that change daily, and dense crowds that attenuate signals unpredictably. Promising granular accuracy in this setting without a measured pilot is irresponsible. The realistic framing is zone-level detection: "this visitor is in the north-east quarter of the hall" rather than "this visitor is at stand B42."
Confirming the fix and preventing recurrence
Quoting manufacturer specifications as deployment guarantees
A data sheet figure is measured in a specific test environment, often with a single beacon and a single receiver at known distances with no obstructions. It is a useful benchmark for comparing hardware, not a prediction of performance in a stocked warehouse or a busy museum. If a proposal presents a manufacturer's accuracy figure without qualification, treat it as a best-case laboratory result, not a deployment promise.
Ignoring the receiving device
Accuracy is not determined by the beacon alone. Different phone models have different Bluetooth antennas, different firmware implementations, and different approaches to RSSI smoothing. An Android device and an iOS device held in the same spot will report different RSSI values from the same beacon. A phone in a pocket will read differently from a phone held in hand. Any accuracy claim that does not acknowledge the receiver as a variable is incomplete.
Confusing consistency with accuracy
A beacon might consistently report an RSSI of -65 dBm at a particular spot on a Tuesday morning with no one else in the building. That consistency does not mean the estimated distance is accurate. It means the conditions were stable. When the room fills with people, the reading at the same physical spot may shift by 10 to 15 dBm, which translates to a significant distance estimate change. Stakeholders should understand that consistency in a controlled test does not guarantee consistency in operation.
Key questions to put to a supplier or integrator
- What accuracy have you measured in a environment similar to ours, and can we see the test data?
- How was that accuracy defined: radial error, zone classification rate, or something else?
- What receiving devices were used in the test?
- How does accuracy change when the space is occupied versus empty?
- What calibration process do you follow on site, and how long does it take?
- What happens to accuracy at the edges of zones, near walls, or in corridors?
Setting expectations with stakeholders
The most effective approach is to define accuracy in terms the business can verify. Rather than saying "the system is accurate to two metres," say "in our pilot, the system correctly identified which zone a visitor was in 94% of the time, based on 500 observations across three device types." That statement is specific, verifiable, and honest about what it measures. It also gives stakeholders a concrete basis for deciding whether that level of performance supports their use case.
If a stakeholder pushes for a single accuracy figure before any site measurement has taken place, the correct response is to explain what variables are unknown and propose a scoped pilot to produce a real number. Offering a guess to satisfy the request only creates a problem later when the deployment cannot meet it.
