Convenience store retailers continue to grapple with how best to respond to the disruption of frictionless checkout. I recently spoke with Crone Consulting’s CEO Richard Crone and Managing Partner Heidi Liebenguth for a podcast on autonomous checkout.
Crone pointed out that Amazon Go is working at a 99.99% confidence level. “If something does fail, from our estimations, somebody is reviewing the video on the backend to see what actually happened and resolving the issue pretty quickly,” he said.
For retailers to compete, they’d need a similar confidence level. “The only way that the artificial intelligence engines can get to that is by interpreting massive amounts of data,” Crone explained.
“If you donate (your) SKU level data, dwell time and checkout data to (autonomous checkout) startups, you’re going to help build a unicorn value in someone else’s business while you’re giving up the very essence, the most prized asset of your merchandising concept,” he said. “And that’s your SKU-level customer data while they’re in the store.”
Liebenguth pointed out that the real value of the machine vision and machine learning used in autonomous checkout isn’t the autonomous checkout itself, but gaining the data behind who the customer is through the check-in process.
“Check-in allows the retailer to personalize the customer journey through the store, not waiting until checkout … but being able to know who they are, communicate with them while they’re in the store and personalize that experience for them,” she said.
It’s important to recognize that Amazon Go is working from a clean slate and not retrofitting frictionless checkout into an existing retail environment.
“They have simplified this (product) mix and created a new merchandising concept. That concept is premium, prepared foods and groceries,” Crone said. “The reason is that this first generation of technology around autonomous checkout can’t read at a high ROC (receiver operating characteristic) curve, or confidence level.”
Amazon has essentially eliminated many of the exceptions found in traditional retail that would pose challenges — from lottery to cigarette sales.
Another approach some retailers are taking is scan and go, but some retailers are seeing shoplifting with this approach.
“Another thing retailers are testing is order ahead, and this — if it’s done right — could be a replacement for self-checkout altogether,” Liebenguth said.
Order ahead can occur in a designated pickup area within the store, allowing c-stores to test machine vision technology in a limited space with a limited SKU selection, ensuring the machine vision isn’t challenged in capturing exactly what is picked up.
The advantage and value of autonomous checkout is in making the customer experience more valuable and increasing visits and basket size.
“Check-in is the area the retailer really wants to focus on because that is where you can get the most bang for your buck as far as getting a higher value out of this whole installation,” Liebenguth said.
Retrofitting a store could be years away, but all retailers today can take steps to connect with customers and make them contactable by introducing their own five-star-rated app and ensuring the data obtained through registration is harvested and used by the c-store.
“Every convenience store retailer needs to exercise their data rights and have a strategy,” Crone said. “Because if you don’t have a strategy, you’re building someone else’s business from that data. …” Launching their own app gives c-stores the ability to compete, to connect with customers and to build a known, enrolled base of customers through a customer relationship model.
Listen to the full podcast here: cstoredecisions.com/2019/09/04/c-store-technology-oracle-podcast-autonomous-checkout/