How accurate are BLE beacons? Four methods, and why results vary

How accurate are BLE beacons? It depends far more on the method than on the beacon. Simple proximity tells you the nearest gateway or room. Signal-strength (RSSI) trilateration typically lands within a few metres. Fingerprinting can do somewhat better in a well-surveyed building. Angle of arrival (AoA), using Bluetooth direction finding, is where published specs claim sub-metre accuracy, with more expensive receivers.

So a vendor quoting "1 m accuracy" and another quoting "room level" may both be honest; they are describing different systems. This guide explains the four methods, why results vary so much between buildings, and what to ask before you buy.

There is also the reverse set-up, where fixed beacons broadcast and a phone or tag listens and works out its own position. That is common for visitor wayfinding with a "blue dot". The same methods and limits apply.

If a gateway sits in each room, the system reports the room. Accuracy is then a question of correctness, not metres: how often is the tag shown in the right room? Signals pass through walls, so a tag near a doorway can flip between neighbouring rooms. Good systems smooth this with rules, such as requiring a stronger signal for several seconds before changing zone.

For finding equipment, counting assets per room or knowing which floor something is on, zone-level presence is usually enough and costs least. The Bluetooth beacon tracking page explains how zone names in your data match rooms on a map.

• RSSI is noisy. The same tag at the same spot can read several decibels apart from one second to the next.
• Reflections. Metal racks, machines and glass bounce signals, so strength no longer tracks distance.
• People. The human body absorbs 2.4 GHz signals. A badge worn on the chest reads differently facing towards or away from a gateway.
• Orientation and mounting. A tag inside a metal toolbox or flat against a cart behaves differently from one in open air.
• Gateway density and geometry. Too few gateways, or gateways in a straight line, give poor positions whatever the maths.
• Broadcast interval. A tag that broadcasts rarely gives fewer samples, so the system must choose between smooth positions and fast updates.
• Filtering choices. Heavy smoothing hides noise but adds lag; a moving forklift appears where it was a few seconds ago.

1. 1
   Which method?

   Proximity, RSSI, fingerprinting or AoA. The answer sets the realistic range before any number is quoted.

2. 2
   Which percentile?

   Average error hides bad cases. Ask for the 50th and 90th percentile, or for room-level correctness as a percentage.

3. 3
   In what building?

   An empty office and a full warehouse give very different results. Ask for figures from a similar site.

4. 4
   At what update rate and battery life?

   Accuracy measured at ten updates a second will not hold when tags broadcast every five seconds to save battery.

5. 5
   How many gateways or locators?

   Ask for the count per floor area, and what changes when racks or walls move.

6. 6
   Does it detect floors?

   Multi-floor sites need the correct floor first. Signals leak through ceilings and atria.

7. 7
   Can we pilot it?

   Run a short test in your hardest area with your own assets, and measure the result yourself.

A vendor demo shows what a system can do. A pilot shows what it does in your building. Keep the test simple and repeatable:

1. 1
   Mark test points

   Choose 20 to 30 spots across the area, including corners, doorways, metal-heavy aisles and busy corridors. Note each spot’s true room and position.

2. 2
   Place real tags on real assets

   Put tags where they will live: on carts, inside toolboxes, on badges worn by people.

3. 3
   Record what the system reports

   Log the reported zone or position at each point for several minutes, at the update rate you plan to use.

4. 4
   Score it two ways

   For zones, the share of readings in the right room. For coordinates, the error at the 50th and 90th percentile.

5. 5
   Repeat at a busy time

   Run the test again during a full shift, when people and goods change the signals.

Write the gateway layout and tag settings down next to the results, so a later change can be compared fairly.

If the job needs tens of centimetres, BLE may not be the right tool. The BLE vs UWB vs Wi-Fi comparison covers the alternatives.

Spatial Map does not compute BLE positions, so it does not set your accuracy. It shows what your gateway software or RTLS engine reports: room presence when a zone name matches a room on the map, or live markers at the coordinates your system sends, on the right floor of a 3D floor plan. Seeing positions against walls and racks is a quick way to judge a pilot: markers inside walls or in the wrong room are easy to spot. See BLE asset tracking and RTLS software.

For visitor-facing "blue dot" navigation, accuracy needs are different again; the airport and mall wayfinding guide discusses them.

Can BLE beacons give centimetre accuracy? Not in typical deployments. Published AoA systems claim sub-metre accuracy; centimetre-level tracking is usually the domain of UWB.

Does adding more beacons improve accuracy? Up to a point. Better geometry and overlap help RSSI methods, but beyond a certain density, noise and reflections dominate.

Why does my tag jump between rooms? Signals pass through walls, so a tag near a boundary can be heard well in two rooms. Smoothing rules and gateway placement reduce this.