Why are there still checkout lines in 2020?!!

Grocery shopping sucks. We all know it. The average trip to the grocery store takes 43 minutes and happens 1.5 times a week. That’s 56 hours spent at the grocery store per year. Absurd. Why does it take so long? Why are there 20 cash registers but only 3 of them are open? Why did I have to spend 10 minutes looking for Honey Bunches of Oats? Why didn’t I know about that sale on Tyson chicken when I bought the Perdue chicken? Common annoyances and inefficiencies such as long checkout lines and hard-to-find items bog us down and people are getting fed up. Creating an experience that solves these problems is known as frictionless retail. I believe a tech-enabled shopping cart is the most effective way to create this user experience. With optical and weight sensors, the cart would automatically detect which items you put in your basket using computer vision (real-time object detection). When you’re done shopping, you simply checkout on the cart itself rather than in a checkout line. Along your shopping experience, you can easily find your items using the cart’s in-store map and navigation. The cart also notifies you of relevant deals and products you may have missed.

Before I elaborate further on the cart’s details, let’s take a step back and look at the challenge we are trying to solve: enabling frictionless retail.

The two biggest stakeholders of this product are end users (shoppers) and customers (grocery stores).

Everyone is busy and nobody wants to waste extra time at the grocery store, but some user groups are particularly impacted by a lengthy and difficult grocery shopping experience:

While the end users are the shoppers themselves, the customers are the grocery stores.

Based on our users’ needs and pain points, we can define some goals.

If this product were a reality, we would want to track a couple of Key Performance Indicators to make sure it’s really accomplishing what we hypothesize it would. The following is a non-exhaustive list.

Individual Unit Metrics:

Shopping Cart Fleet Metrics:

There are a number of business models that could be applied to this product. Grocery stores could be asked to buy the carts up front, the carts could remain under the ownership of the manufacturer while the store pays a monthly/yearly fee, different features can have different prices, etc.

Is it worth it for the stores? Let’s find the breakeven time.

Cost of Cart Estimate (3 year lifespan): $1500, $500/year

Cashier Average Salary (full-time): $27,413

Number of Customers Served per hour:

Traditional Checkout Cost per Customer = [(salary) / (number of hours worked per year)] / (number of customers served per hour) = $0.913

Tech-Cart Cost per Customer = [(cost per year) / (hours operational per year)] / (number of customers served per hour) = $0.048

Because the tech-carts are able to efficiently and effectively emulate the traditional process, grocery stores reduce their costs greatly, freeing their cashier clerks to perform other duties like restocking or assisting shoppers. The above calculation on the financial benefit to stores doesn’t even take into account the potential increase in sales due to more customers coming to their store (87% of respondents said that they would actively prefer a store with frictionless checkout over one with traditional checkout).

There are some efforts in the tech industry to create a frictionless retail experience, but all are either financially or technologically infeasible.

Amazon Go: The most visible example is Amazon Go, the store where users can simply walk in, scan their personal QR code, pick up items, and walk out. Cameras and sensors all around the store automatically detect what each person has purchased and charges your Amazon account. After visiting an Amazon Go store, I realized that this was not the right solution. The amount of camera and sensor hardware was astonishing, lining every square foot of the ceiling. The store was tiny — Go stores are 1,800–2,100 sq. feet compared to the 40,000 sq. feet of the average American grocery store — and yet the hardware alone is slightly above $1 million per Go store. Experts estimate that the breakeven time of a Go store is over 2 years, making it financially infeasible!

Camera-Only: Another effort in frictionless retail is a similar system to Amazon Go’s, but one that only uses a fraction of the cameras, mounted around the store above shoppers. NCR Corporation is at the forefront of this strategy. When I worked in the lab at NCR that is responsible for the R&D of this initiative, I saw firsthand that it was not accurate and robust enough to be a production-ready solution for the Krogers and Walmarts of the world. The computer vision technology just isn’t ready for that level of precision at that distance and it doesn’t account for weight-sensitive items like fresh produce.

CV Kiosks: Finally, there are Computer Vision Checkout Kiosks such as the one made by Mashgin. These are basically self-checkouts that use computer vision instead of barcodes to scan your items. I’ve used this product and there’s a glaring problem — this isn’t frictionless! You still have to unload all of your items and checkout, and it takes even longer than a traditional self checkout or cashier.

One indirect competitor would be grocery delivery services. More and more users (shoppers) are switching to grocery delivery, largely because of the current frustrations with in-store shopping. Grocery delivery comes with its own myriad of downsides: hefty pricing mark-ups and fees, wrong/missing items, requires you to schedule ahead of time, doesn’t allow you to select your own produce and meat, etc.

High Cost: In my user interviews and research, I found that grocery delivery costs between 15%-28% (depending on the store and delivery service) more than the exact same purchases made in-store.

Low Flexibility: A common qualitative remark I heard from users was that grocery delivery “requires way too much planning and scheduling. You have to be at home when the groceries get there and set the delivery window far in advance.” In contrast, showing up to the store whenever you want allows for greater flexibility and users are accustomed to that after a lifetime of in-store shopping.

Less Control: 42% of consumers say they prefer the ability to choose their own products, like produce and meat, and delivery doesn’t allow you to do that.

While grocery delivery is growing as part of the macro-trend of delivery services, it is clear that in-store grocery shopping is here to stay if its key user experience issues are solved.

The rise of in-store shopping frustration along with the concurrent rise of grocery delivery is threatening the existence of traditional grocery stores. They need to address the serious problems in their shopper/user experience by enabling a frictionless retail experience. The frictionless solutions currently in development are either not financially feasible (Amazon Go), not technologically feasible (Overhead Camera-only), or not frictionless (CV Kiosks). A cart-based solution solves all of these problems. By concentrating the technology in the shopping carts rather than around the entire store (like Amazon Go), the tech-cart solution is financially feasible and scalable to any sized store (even Costco warehouses). By using camera sensors that are localized to the cart itself rather than above shoppers, the computer vision aspect becomes technologically feasible. By eliminating the need for checking out in the traditional manner, the tech-enabled carts prove to be truly frictionless.

The tech-cart (or as I have taken to calling it: the SwiftCart) just makes sense. It makes sense for shoppers and it makes sense for stores. When a product makes this much sense, I believe it should exist.

Related posts:

She may be captivated by a composition book. It is suitable for writing without distractions. In her delicate mind, she’s free to write whatever she wants. Give her the feeling that she may have that…

In 2016 the US had an election. They seemed to have the choice between a woman and a buffoon. It also seemed that the buffoon was trying to make people believe that the woman was fraudulent, or at…

Recently I took a dive into learning Version Control systems(VCs). As the name suggests, systems that keep track of the growth of files:-from the very basic empty file to a fully edited file. My…