SMB Innovation Lab: Face Recognition with in-Store Analytics
Within every footstep is an opportunity to grow your in-store strategy
Click-throughs, bounce rates, segmentation and conversion funnels are common and essential metrics for any online store. Brands which do not employ these to enhance the experience will miss out on targeted content, display optimization and personalized recommendations which account for over 20% of an online-stores revenue.
When it comes to bricks and mortar these tools are sorely missing. Sessions, cookies and preferences just don’t exist in the real world. Until now.
We take the latest artificial intelligence breakthroughs in the field of vision-based Human Computer Interaction (HCI) to change the game. With In-Store Analytics, SMB Retailers can leverage the power of digital footprints to apply the same tools as their online counterparts.
In-Store Analytics is an app prototype we built in the Innovation Lab. It is loosely coupled into SAP Business One and can be deployed on any cloud server… such as the SAP Cloud Platform.
To install the concept yourself, understand how it works and use it as a basis for any Face Recognition project you’re working on, the source code is available in our GitHub account here. It is released AS-IS with no commitment or support – though we’d love to hear if you’re working on something. Read on for details of how to set it up.
Face Recognition Fundamentals
In-Store analytics is built on a foundation of face recognition. It uses the feed from an in-store camera, something which is ubiquitous in modern retail. We enhance the real-time data coming from this feed to identify when a face crosses a chosen zone within the camera’s field of view.
Once a face is recognized, it is converted into a digital ‘Face ID’. We do not store face images, rather a hash-like matrix which acts as a thumbprint. This is compared against all the Face IDs already stored in the customer’s database for a match. If no match it is stored as an anonymous visitor, in the same way as a cookie on a website.
In-Store analytics identifies visual clues about the person based on their face. It estimates gender, age and sentiment, or mood. In our case this ranges from Angry to Very Happy
In-Store Analytics in Action
Building Customer Connections:
Our prototype takes place within a wine store. It is designed for a store manager or marketer, or the role(s) responsible for understanding or connecting better with customers. In our case the showcase is a screen, though it could be part of the POS or on a separate display… or in the corner of a pair of smart-glasses.
It is loosely coupled into SAP Business one and so is built around suggesting the right products to customers as they enter.
The People tab shows those face who have recently entered the store. Clicking on a face opens their profile on the right. If a person enters the green brackets in the center of the screen, their profile will automatically appear on the right:
Three types of profile views:
- Known customers who have previously signed up for membership are connected to their customer accounts. Their profile information along with product preferences and relevant discounts are applied.
- Repeat customers are recognized and incentives applied to aid with sales conversion. In our example an individual has visited the store more than five times but hasn’t become a member. We apply a discount to the recommended products.
- Anonymous customers receive a product selection which helps break the ice in communication. This can be based on a machine learning algorithm which takes several factors including facial clues and environment conditions. Did you know that happy people drink more red wine!?
Gathering store-wide insights
In-Store analytics builds deep insights into the types of people who visit over time. This builds a digital dashboard of otherwise analog information in both the Visits and Persona tabs.
- Visitor Frequency and average session times
- Bounce rate – the ratio of visitors who leave with 30 seconds of arrival
- Correlations between revenue and visitor volumes
- Demographic analysis across age and gender
- Customer profile types – anonymous/repeat/member
- Sentiment analysis including mood-over-time trends
If you are interesting in running in-Store Analytics and are familiar with app development, the prototype is available in our git-hub account AS-IS (we offer not commitment or support).
Installation instructions and access to the open-source code: in our github
About the SAP SMB Innovation Lab
As part of SAP’s SMB Group based in Shanghai, China, we mold the hottest technology trends into inspiring innovations. We act as an internal startup poised on the shoulders of a world market leader to get our hands dirty in defining and creating new ways of getting business done.
More about Face Recognition: Your Face is the Key to Unlocking Multichannel Retail
We’re hungry for feedback and always happy to talk. If you’re working on a topic to do with Face Recognition, have an innovation idea or want to know more please do contact us: firstname.lastname@example.org
In-Store Analytics and the SAP Digital Core
|True traffic measurement|
|Learn from consumer behavior to improve conversion rates and transaction size. Measure the entire-funnel from walk-by through to completing a sale. Automatically exclude store staff from traffic counting and other KPI metrics. Compare your performance against your industry on a local, regional or national level.|
|Measure & Optimize for growth||Optimize staff|
|Ensure product placement and displays drive exposure and engagement. Validate store layout and campaigns. Test and measure new concepts such as product placement, store design and layout.||Identify traffic cycles across the day, week, or year. Align staff and customer-to-staff ratio to traffic power hours by integrating workforce management with traffic counting data|
|Engage in real time||Personalization|
|Through the mobile app, view real-time data and video on the go. Get real time alerts if a VIP enters your store.||Display recommendations based on previous activity and preferences, as part of your omni-channel marketing strategy.|