Retail is going through a massive transformation in multiple fronts. Online Stores are racing ahead with several advantages compared to Offline stores. The data collected by the online stores about the buyers is becoming a huge advantage to target products and to come up with merchandise that is likely to address the general set of buyers visiting the online store.
To compete, offline stores now are going through digital transformation addressing multiple segments with mobile applications etc.
The next phase is to get buyer patterns, similar to the online store. Offline stores have an advantage in this respect that they have real customers but there is a technology gap in a system solution to address this easily. Gyrus presents a solution that addresses this requirement by fusing information from Camera, Point of Sale (POS) and Customer Location (LOC) to enable Smart Retail Stores.
Camera based Analytics
With the advancement of vision algorithms, cameras and the associated AI hardware provide lot of insights about the environment. Object detection, face recognition, sentiment analysis, demographic classification algorithms matured and are being used in real-life deployments.
Stores can have cameras at the entrance and exit or can be in every aisle. Cameras in every aisle provide deep insights and very granular information, including sentiment analysis. However it will be very expensive to install, and maintain such setup (as every aisle will need to have cameras with their orientation monitored) The maintenance expense will be more in the long run than the installation cost. The multiple cameras also requires hardware to process the multiple streams such as GPU’s, Servers etc. Cameras in every aisle can be intrusive as well.
Alternatively, Cameras at entrance and Exit can be used to identify
- Demographics of the visitor
- Rewards customer
- Frequent visitor
- Cross check POS with checked out items.
As an added advantage, cameras can provide security and surveillance. Camera software can alert if someone enters (or is walking outside) the store with a weapon.
Point of Sale (POS)
POS systems should be able to provide information regarding the products purchased, discounts/coupons applied, loyalty rewards etc. The backend software can correlate them back to the categories of the products.
Wi-Fi technology introduced Fine Time Measurement (FTM) for getting more accurate location measurements. Advanced methods are layered on top such as Channel Estimate based Time of Flight measurements and Trilateration Technologies to have much improved location metrics. Today it is conceivable to have decimeter-level (20 cm) accurate location of the user devices. The previous generation WiFi-APs used BLE, or antenna arrays for location, and dense deployment making it very expensive, where these new methods can be deployed with software running on standard WiFi AP’s.
Fusing the three vectors
Taking in data from the three technologies, offline stores can be provided very valuable insights at par with online stores such as heat maps below.
Based on the time spent by user at each product location, their demographics, and the reconciliation with the POS system, one can provide –
- Foot Traffic to Store Traffic rate
- Time Spent by per Aisle
- Time Spent per product section
- Conversion Ratio
- Conversion to Time Spent
- Business Lost Ratio
- Business Lost to Time Spent
- Marketability of the Merchandise
- Product placement effectiveness
- Effectiveness of discounts
- Time spent in Mark-Down section
- Conversion ratio in the Mark-Down section
- Demographic wise statistics of Business gained / lost
- Effectiveness of loyalty program
- VIP Customer Identification
- Effectiveness of Digital Signage
- Customer-ID to online ID mapping
- Targeting advertisements
- Customer Heatmaps
- Queue Monitoring
- Zone Analysis
- People Counting
- Shopper Flow
The online retail companies are threatening offline stores with respect to inventory level, margin of business, requirement for less labor etc. The latest battle is with respect to data about buyer profiles and patterns. The online stores develop rich buyer models and personas to target advertisements and provide custom merchandise to lure them to buying. Offline stores can have similar information and can do better targeting and with new technologies as presented in this blog. Same level or better information can be available to the offline stores to help them compete better. A mesh of Cameras everywhere is expensive, very intrusive and might thwart some customers. Intelligently fusing the Wi-Fi sensing, POS systems with Cameras just at the entrance and exits, intelligence can be brought to offline stores at par with the online stores.
Gyrus ML/AI algorithms can be used to fuse the data and also to derive all the insights presented above.