The Checklist for Superior Frictionless Shopping

Erez Arbel

VP Product

11 July 2021

As the grocery retail industry moves further toward adopting in-store automated solutions for faster, more convenient shopping experiences, smart, checkout-free technologies are seen as the next holy grail. By 2025, smart checkout technologies are set to process nearly $400 billion in transactions, up from $2 billion in 2020, according to a recent Juniper Research report mapping the future outlook in retail. Similarly, retailers are expected to pour some $23 billion into artificial intelligence solutions in the next four years, driven by a need to upgrade operations and deliver a compelling, efficient shopping experience.

At Trigo, we developed a system that enables easy, automated, checkout-free grocery shopping in existing retail stores where shoppers walk in, select their items, and simply walk out. We work with Tier-1 grocery companies to retrofit existing spaces that allow for this supermarket-of-the-future type of experience, complete with digital tallying of items and online payments and receipts via a mobile application.

This, we believe, is the future of retail and we have worked hard to develop the industry-leading capabilities to make this shopping experience a reality.

How do we ensure a truly frictionless, superior grab-and-go shopping experience? Here are some of the key elements we worked to hone:

Intuitive onboarding and in-store orientation for shoppers

Introducing new systems and processes, especially into more traditional industries, can be fraught with worries and concerns. We at Trigo developed an onboarding process for shoppers that is intuitive and keeps things simple, while also navigating the differences between retailers (in different countries, for example).

A simple set-up would have shoppers input their payment information and other preferences and then provide information and instructions about in-store processes like following along as items are added, removing already tallied items, and requesting refunds or changes where necessary.

Shoppers will also be instructed on small changes to in-store behavior like how to purchase items from the deli or bakery section which can vary by weight and product, or handle age-restricted items such as wine and liquor.

Trigo has put a lot of effort into creating onboarding that makes sense across the board and is easy to adopt, regardless of the differences in layout and products per retailer.

A natural shopping experience

As the adoption of smart solutions gains pace, retailers are evaluating different types of technologies like self-checkout and scan-and-go. In practice, these can create more work and require greater effort from the shopper, now tasked with finding barcodes and making sure the items are processed correctly. 

It can be a frustrating journey, and in-store employees must often intervene when items are not scanned or weighed accurately by the system or if it inexplicably refuses to process a working card.

Trigo’s solution is truly frictionless and seamless, and involves no active process with shoppers beyond the initial set-up. This allows for a natural, intuitive, and pleasant shopping experience that crucially skips *any* checkout line.

A high level of accuracy

Trigo’s technology is based on 3D mapping, computer vision, and machine learning combined with off-the-shelf hardware such as cameras and shelf sensors. The ceiling-based cameras record a given store’s interior from different angles numerous times per second to generate a 3D model of the position of the people and items in the shop. We then apply complex proprietary algorithms to capture anonymized data on shoppers’ movements and product choices, automatically monitoring the complex real-life, in-store interactions.

Our system has to be highly accurate for this to work. Retailers must feel confident that we can correctly capture and process all the natural events that occur in the store. From people changing their minds and putting back those brussels sprouts they ambitiously picked up for dinner, or grabbing the wrong type of bread and dropping it in the pasta aisle (potentially stashing it among the selection of vermicelli). Or, if a child, for example, picks up a small bag of gummy bears and tries to surreptitiously hold onto it while their parent is distracted.

To achieve high-level accuracy even in cases where items or products may not be in full view or in the wrong location, the system combines pressure sensors with vision prediction, ensuring retailers can identify all items. The different data streams are integrated with unique AI sensor fusion algorithms that allow the system to deal with unexpected occurrences and spot non-visible interactions, such as people trying to take items without being noticed.

In fact, our technology helps grocery retailers significantly reduce shrinkage, or loss of inventory/money due to shoplifting, employee theft, cashier errors, administrative mistakes, vendor fraud, and the like. According to a 2020 survey by the US National Retail Federation, retail shrink is at an all-time high, accounting for 1.62% of the average retailer’s bottom line. With Trigo’s tech, we’ve found that this percentage rate could be brought down to close to 0.

Since Trigo’s systems account for different sets of scenarios and variables, we can deliver an accurate, efficient, and cost-effective experience for retailers and shoppers alike.

Support for a wide range of products

Since Trigo specializes in retrofitting existing stores with its unique technology,  ensuring retailers keep their distinctive character and layout, the system supports a wide range of product lines. 

Customers can shop for their fruits and vegetables, pick up deli meats or cheese by weight, select pounds of fish or fresh bread, and even go to the service areas and order a cappuccino to-go and ready-made salads and dishes – all with the same frictionless checkout tech and without having to pay at different stages.

The ability to scale

A key technology barrier in scaling is the training of AI models on many new products depending on the retailer and the region. AI models need thousands of annotated examples to recognize items and since supermarkets usually offer thousands of products, deep learning algorithms require tens of millions of annotated examples. 

Trigo developed an automatic annotation data pipeline, designed for training the neural networks with a constant stream of generated examples. This cutting-edge technique enables Trigo to generate data for new retailers and solve the challenge of adding goods to the store every day. So retailers don’t have to worry about the coverage of new items or products on a daily basis or for specific holidays or seasonal events.

Lastly, in addition to the in-store core system, Trigo enables the distribution and implementation of the technology at scale in multiple stores. And at a relatively low cost. So retailers can deploy Trigo’s tech across their chain stores with minimal disruption. 

Trigo is constantly at work developing and improving our world-class capabilities and setting new industry standards that allow retailers to take their operations forward while offering easy, efficient shopping experiences to their customers.

Our retail-ready tech is already deployed at a number of stores in Europe as we work with more grocery giants across the world to deliver next-generation shopping.