TraxRetail

 

“One single source of shelf truth”

Founded: 2010, Singapore

Category: Artificial Intelligence

Primary office: Singapore

Core technical team:Tel Aviv, Israel

Status: Private

Employees: 251-500

Amount raised:USD$386.9 million(9 rounds – July 2019)

OVERVIEW

  • Scaling strategy is a combination of crowdsourcing (visits to stores) and advanced computing engines (data validity, computer vision, image recognition) along with robotics – Trax recognizes 1 billion SKUs per month
  • Trax offers various products and services that lead to an understanding of how clients’ consumer products look, perform and compete on the retail shelf
  • Distinct competences include computer vision applied to retail
  • Using computer vision, deep learning and other AI technologies to enhance how clients’ products look, perform and compete on store shelves

PERFORMANCE METRICS

  • Valuation: USD$1.1 billion (2019)
  • Estimated Revenue: USD $73 million (2019)

ACHIEVEMENTS

  • Numerous industry firsts including:
    • Introduction of real-time photo recognition analysis to retail
    • Scale above 500,000 store visits per month
    • Deliver KPIs to the salesperson in the store
    • Brought retail image recognition to mobile smartphones
    • Recognize nearly 1 billion SKUs per month
  • Deloitte Technology FAST 50 Award (2016)
  • Frost and Sullivan’s Product Innovation Award (2017)

Sells

  • Trax retail execution – execution optimization across stores and field teams – helps consumer goods companies win at the shelf with a comprehensive real-time view of store and field performance across all retail channels
  • Trax retail snapshot – quantifies in-store merchandizing and promotions at scale and speed; results in improvements to execution, optimize new launches of focus retail efforts on highest value activities
  • Trax dynamic merchandising – AI driven, dynamic merchandizing service
  • Shelf intelligence suite by Trax and Neilsen – integrated solution for continuously measuring, optimizing and activating store strategies
  • Shelf pulse and shelf blueprint by Trax and Neilsen – pulse is a syndicated data delivering visualization of sales and store conditions; Blueprint leverages advanced retail data science models to enable optimal store activation, macro space planning, and shelf optimization designs
  • Trax retail advisory – strategic insights tailored for brand and category growth
  • Trax retail watch – real-time monitoring and store management
  • Various solutions for store operations and category management

Channels

  • Partnerships with leading innovators of IoT technologies (including fixed cameras and robotics)
  • Partnered with Snooper, which pays 75,000 shoppers, students and retirees to take photos of products on supermarket shelves and upload them through an app (Crowd sourcing)
  • Partners with Google Cloud Platform, IRI (for better execution and analytics) and Kantar Group (optimize category management and product assortment)
  • Shared solutions between Trax and Neilsen (see above)
  • Academic partners include MIT, Bar-Ilan and Ben-Gurion University of the Negev
  • Trade shows, international partners (e.g., access into China)

Competencies

  • Application of computer vision to retail (particularly shelf space)
  • Deep learning, other AI techniques within a cloud-based process for rapid and accurate image processing
  • Technical – algorithm engineering, computer science, machine learning, geometry and computer vision

Distinct AI Features

Type

  • Computer vision, deep learning

  AI use

  • AI and crowdsourcing are used to deliver fast, accurate and consistent means for customers (beneficiaries) to optimize how their products are executed on the shelf – primarily based on proprietary computer vision algorithms
  • Trax is involved in academic research (areas include algorithm engineering, computer science, machine learning, geometry and computer vision)

AI useRate of return on customer’s investment to make AI work

Immediate:

  • Greater efficiency and effectiveness of consumer goods leading to measurable increases in sales
  • Accurate and scalable recognition of shelf products leading to enhanced situational awareness
  • Data driven insights allowing for role-specific (e.g., management or onsite sales representative) actionable insights at speed and scale

Long term:

  • Optimization of client products on the retail shelf – fast, accurately and consistently

Databases

  • Largest retail image repository (database)

Quantum Computing

  • No explicit mention of quantum computing in Trax collateral. Generically, quantum computing is expected to enable artificial intelligence by, for example, handling huge amounts of data, building better models and more accurate algorithms, and using multiple datasets (integrating datasets more quickly

Resources

Assets

  • Largest retail image repository
  • Deep learning algorithms for retail
  • Proprietary computer vision algorithms
  • Cloud-based advanced engines for extracting insights from images – e.g., stitching shelf images and acquiring correct SKUs; location and size of each SKU identified; generated data used to create scene and session facts
  • Computer vision platform
  • Documented use cases (e.g., Molson Coors – required higher visibility into store sales and in-store conditions)

Processes

  • Well-developed process from crowdsourcing photographs from stores, uploaded to Trax cloud and processed to obtain insights (that can be fed to management or instore staff, sales representatives)
  • Digitizes every product in every aisle

Priorities

  • Acquisition of complementary assets
  • IPO
  • Incorporate Virtual Reality into products and solutions (using Apple’s ARKit)

References

  1. Trax.2020.https://traxretail.com. Accessed 21 August 2020.
  2. Owler.2020. https://www.owler.com/company/traxretail. Accessed 21 August 2020
  3. Bloomberg.2019. https://www.bloomberg.com/news/articles/2019-05-27/singapore-startup-trax-is-raising-funds-at-1-1-billion-value. Accessed 21 August 2020

Contributors

  • Dan Craigen