Bluetooth BLE Electronic Measurement and Associate Tracking
Bluetooth BLE Electronic Measurement and Associate Tracking
By Gary Angel|
December 14, 2020
Bluetooth Badges are one of the go to techniques for Associate tracking. They cover a wide-range of use-cases and are inexpensive to deploy, making them an attractive option of most forms of labor tracking and people-process optimization.
How exactly does the technology work? Bluetooth tracking relies on each Associate being issues a Bluetooth badge. These are small (common form-factors include keycards or belt-clipped thumb-drive sized devices), battery operated, and very inexpensive. The badges send out an electronic ping on a time basis – as much as three times a second. To use these signals, you need a BLE sensor fairly close to any area you want tracking in. And fairly close, means just that. Part of what’s good about BLE is that it’s a lower power signal than WiFi. That low power means the signal travels less distance and is less confusing to geolocate, but the trade-off is a need for more sensors to detect the signals. You need quite a few sensors to pick up signals throughout an area and ensure that you don’t have blind-spots. Fortunately, the sensors, like the badges, are inexpensive and easy to install, meaning total deployment costs for BLE are low. And if your doing electronic shopper tracking, adding Bluetooth sensing to the WiFi sensors is essentially free.
Each Bluetooth badge is uniquely identifiable with a stable MAC address. So if it’s assigned to an individual, you get consistent, identifiable tracking on a continuous (3FPS) basis. The FPS rate isn’t as good as camera (typically at 15 or 30 FPS), but it’s more than adequate for ANY situation we’ve ever encountered and it’s far cheaper than camera to deploy.
Given that, what can you do with Bluetooth tracking and what can’t you do?
When we first started Digital Mortar our focus was entirely on tracking shoppers – but it quickly became clear that tracking Associates was almost equally important. In our original WiFi based measurement systems, we would routinely pick up a variety of Associate and Store Devices. Even without a list of those devices, it was pretty easy with basic ML to identify them as Associates based on their time and movement patterns. But this kind of tracking was always limited because many Associates don’t carry a phone or tablet, the positional accuracy is poor, the ping rates are slow and random (so that you could easily miss when a shopper and Associate were nearby), and it was impossible to identify individual associates.
Bluetooth tracking solves most of these problems. BLE badges can be assigned to individual Associates. They are inexpensive enough to deploy universally with every Associate in the store. They ping constantly so measurement is continuous. And because they are lower-power than WiFi, the positional accuracy (at least if you have enough receiving sensors) is better.
This makes Bluetooth tracking quite good for higher-level positional tracking of Associates to understand how they spend time. Where they are on the floor. How often they have to go the back-room. And what the ratio of shoppers to Associates is in an area. BLE is accurate enough for section level positioning – so it’s perfect for profiling how Associate time gets allocated by area and measuring intraday STARs. That’s very useful stuff.
What can’t you do well with Bluetooth measurement?
The biggest “can’t” comes from the lack of positional accuracy. Although BLE signals don’t travel like Wifi, they still don’t allow for super accurate positioning. In a dense sensor environment, you’ll probably be able to get positioning in the 10-20 foot accuracy range. That’s not bad, but it’s not really good enough for people-process analytics.
With the onset of Covid, a lot of companies have had to retool their processes – both back and front of office. That’s raised a lot of interest in tracking those people processes in more detail. To better support this kind of analysis, we even built out a full session replay capability in our main DM1 platform. To support capabilities like that, we’ve found that people process measurement has to be positionally exact. You need to really see how a person moved or the tool doesn’t work. With electronic positioning, a person will jump around in a chaotic manner and all the fine-grained movement detail is completely obscured.
To manage that, our people-process tracking implementations rely primarily on camera. These implementations are undeniably cool. They involve comprehensive coverage of the work area and they give highly detailed information about how processes were executed. Even here, though, we’ve found there is a role for Bluetooth tracking – just not as the primary measurement method.
Despite the level of detail that 30 fps camera provides, there are a few gaps. First, these back-offices are crowded environments. Cameras get confused by boxes moving on conveyor belts and huge carts being wheeled around. Worse, employees are often going in and out of doors, into offices, and into freezers or storage areas and camera loses track of them. That’s not so bad, except that when they emerge, it doesn’t know it’s the same Associate. For businesses doing process analysis, that’s a significant problem. We can sometimes write logic to handle that, stitching the two sessions together, but depending on how the environment works, brute force stitching isn’t always workable. This is a problem Bluetooth Associate tracking can help solve.
By positioning a sensor over doors and in obscured areas, we can know with high confidence when an Associate went in or out of the door and when they came back. And because that Associate is identified, we can keep their process trail intact.
So the way we use BLE on people-process measurement is to connect blind-spots in the camera. When someone enters a door, the camera records the exact second. It then consults the BLE sensor data and uses a door-crossing model to identify the tag that just passed through the door the camera is monitoring. Since the BLE tag is identifiable, it allows us to join that session to the correct Associate across visual break-points – a capability that turns out to be extremely valuable.
It’s a small but critical role in camera-based people-process analytics.
- In-Store Shopper Metrics: Dependence and Independence
- Conversion per Opportunity: Combining PoS and Shopper Journey Data
- The Role of LiDAR in People-Measurement and Shopper Analytics
- Tying Point-of-Sale (PoS) data to Shopper Journey Data
- Building a Metric Framework for In-Store Shopper Journey Measurement