Overview
Telstra’s insightful innovation
Telstra Retail is the largest customer facing channel for customers across Australia to buy products and services from Telstra. With nearly 300 stores distributed across the nation, Telstra Retail helps Australians access the nation’s best mobile network as well as buy additional hardware and services to make the most of their connection at home or in their business, wherever they may roam.
Telstra Retail wanted to find an automated solution for accurately analysing foot traffic in its network of stores across Australia. Working with Telstra Spatial Insights, new AI-powered video analysis technology was piloted to develop a simple to use insights system that also protects the privacy of all store visitors. The new system offers greatly enhanced accuracy, compatibility with a wide range of camera systems, and holds potential for a wide range of insights to support retail business decision making.
-
0number of stores using TSI capturing data
-
0%+improvement in accuracy of foot traffic analysis from previous solution
-
0%+accuracy of total footfall data
Challenge
Building data-rich insights for physical retail environments
Operating physical stores presents significant management challenges compared to the digital world. Online retail offers direct measurement of website traffic, user browsing behaviours, and a wealth of clear data insights. Offline, running a retail store has far less tangible data to work with, making it more challenging to manage the location of stores, staffing levels and plan store layouts within stores to maximise sales and marketing performance.
At Telstra Retail, there was a desire to find a way to apply technology to the problem of understanding visitors to physical stores. For traffic counting, traditional sensors were already in use to capture a broad picture of foot traffic that offered limited utility.
Telstra Retail was already working with another Telstra team on an internally developed security camera system in its stores. Exploring the potential to use camera systems for more detailed traffic analysis seemed to hold great potential. An advanced solution would ideally observe traffic into a store but also movement around the store. If video analytics could be applied to this challenge in a way that also protects the privacy of customers, there would be tremendous opportunities for new insights for store planning and management.
The foot traffic system we already had in place was costing a lot of money to sustain and was only directionally accurate. Footfall is a key measure of customer demand in our retail stores, and this information is critical in managing profitability and customer experience. It was a question of how we could make this data more accurate to support improvements in CX and investment for continued growth in our Retail business.
Robert Ibsen
Retail Channel Operations Manager, Telstra.
Solution
A.I. insights on a privacy-first platform
Telstra Spatial Insights (TSI) is dedicated to creating data-driven analytics solutions to deliver near real-time intelligence in real world environments. Partnering with Telstra Retail, TSI piloted a new in-house AI-powered video analytics solution to initially deliver an improved traffic count system and then look to expand the insights through software updates.
The TSI AI video analytics solution was able to tap into camera feeds from existing store CCTV equipment, offering a camera agnostic approach that supports almost any high-definition IP security cameras. This minimised cost and installation requirements, making it faster and cheaper to get each store’s analytics online.
From the outset, the video analytics system was designed to protect customer privacy. The analytics hardware features edge compute technology so that the AI analysis takes place in-store, ensuring that no video would need to be uploaded to central servers. This offered both a cost saving on data transfers and a privacy feature by limiting where video was accessible. The system also included identity masking through blurring the footage flowing through the system, and used a real-time process that did not save the video footage.
Approach
From pilot to national scale at rapid pace
While the new Telstra Spatial Insights video analytics solution largely worked with existing in-store cameras, adjustments were made to better align some camera positions for analytics purposes. For foot traffic insights, optimal positioning was often to take a top-down view on areas of the store – offering greater clarity over movement in, through, and out of stores.
The pilot phase also allowed TSI to refine the tools used by stores to define monitoring zones in the system. The aim was to ensure the tools for defining areas of interest – called ‘triplines’ and ‘polygons’ – required minimal knowledge to setup. Along with improved foot traffic accuracy, polygon tools also track insights on areas of the store a customer is moving through. This can help understand how long customers are spending in certain sections of the store, allowing for data-driven decisions on store layout improvements or monetisation of space.
After the first rollout to over 100 stores, approximately 200 additional stores around Australia were added to the roll out when Telstra brought all branded stores under internal control. With stores spread across Australia, from major cities, to country towns, and to truly remote locations like Port Hedland and the Whitsundays, Telstra Spatial Insights knew the system had to be able to operate remotely with confidence.
A diligent approach was essential through testing, installation processes and remote management techniques to minimise faults and avoid costs due to error. In fact, the Telstra Spatial Insights video analytics system design is robust enough that it is proving suitable for many tasks beyond retail environments.
Impact
Planograms, demographics – and beyond
The new Telstra Spatial Insights AI-powered video analytics in Telstra Retail stores has delivered immediate benefits.
Within a matter of weeks, the initial proof of concept improved accuracy of foot traffic analysis to over 95% (a >20% improvement in accuracy compared to the original solution)
The AI-powered system also holds potential to expand its capabilities beyond simple traffic counts. By developing new AI models and upgrading the system remotely, features such as demographic analysis and customer interactions with specific products on shelves become possible. Such enhancements hold potential to make data-driven decisions about in-store experiences, including marketing messages, product placement and store layout planning.
Related solutions and capabilities
Telstra Spatial Insights
Telstra Spatial Insight is a data driven analytics solution that generates insights on how open spaces, built environments, transport links and other government assets are utilised.
Intelligent Retail
Optimise your in-store and ecommerce operations and enhance the customer experience with Telstra's intelligent retail solutions.
Innovations and Experience
Experience the future of businesss innovations with Telstra and explore advancements for enterprise growth.
