To optimize physical stores, we need tracking data. In evaluating people tracking solutions, we learn the differences between competing and complementing technologies. There are three categories to compare.
The first category is sensors. These include video, thermal, and time of flight.
The key point about sensors is that they cover a specific area. The goal of the sensor is to capture ALL behaviors within the Field of View. The data is of all objects, and all the population.
This is important, for example, in Door-Counting. The number of Visitors carries demand, scheduling, and sales conversion metrics. The business value of those metrics requires that the raw data will be accurate.
There are different sensor technologies and solution providers. And they compete on costs, service, and accuracy. In sensors, accuracy of detection is paramount.
The sensor must capture all the objects, and all behaviors.
The current trend is toward object tracking, which allows for more data. The insights on Time-Based metrics and Buying Groups are required for InStore Analytics.
(Dynamic Location) Devices
People Tracking from devices started with the introduction of smart phones. Today, any wireless technology, from product tagging to wearable, provides tracking data.
The key point about devices is that you are not tracking the object, but rather the device. This is a crucial point. For example, WiFi Tracking is about Device ID. Saying “ID is a single person” is an assumption, not empiric data.
If an individual does not carry a phone, the WiFi feature is not active, or has more than 1 phone, the data is skewed. Those situations are much more common than you may assume. As a result, the analytics has different business value. With devices, the focus is on tracking the end-to-end customer journey.
The advantage of devices is how easy they are to track. Their wireless nature makes them an attractive cost and convenience source of data.
The business value is the ability to capture the individual journey, of either a person or a product. I call this concept Tag & Track.
The challenge is in the nature of the data. One aspect is that each signal is a sample from a single device. The quality of the data is statistical. In such, we must know more about data consistency and quality of the sample.
Wireless technologies are coming in many forms, such as WiFi, GPS, BLE, NFC, and RFID. All together, they play a critical role in the formation of the IoT Universe.
For People Tracking, the data from the devices is complementary to the output of the sensors. If you have data from devices and sensors, you can get a much better picture of behaviors in the physical store.
Quality of Data
The third component in assessing a solution is the quality of the data. Regardless of the solution provider, the data quality has certain dependencies.
The first factor refers to the hardware and software architecture. The obvious question is which part does the bulk of the computation. The common scenario is a smart sensor that calculates the metrics inside the store. And the output is data fields. Tomorrow, the scenario may be a video stream captured by a high-end analytical engine.
The second factor is a derivative of hardware/software scenarios. This is about Area versus Path Analytics. If we look at an area, the data describes objects detected in that particular location. If we are looking at a path, the data is about specific objects.
Detection of objects is critical to location analytics. Tracking of the same object across many locations is the core of path analytics. In other words, the business value weights what is more important in the solution.
The third point about data is what I call “Good Enough Accuracy”. This is such an important topic that we have a course on accuracy. Here’s we will breeze on the accuracy requirements.
Good Enough Accuracy depends on the goals for tracking.
There are different levels of accuracy for detection, recognition, and tracking of objects. Each component is important but some are more than others for the business goals.
Another key factor in accuracy is the audits. How we audit, and the period of the audit, has a direct impact on the levels of accuracy. The Accuracy Rate is easy to sell, but the nuances of the data have a direct impact on the quality of the data.
Good Enough Accuracy is a key to the business value of the People Tracking Solution. And once you know your business objective, the choice between sensors or devices is obvious.