CHICAGO -- Acknowledging the complexity of artificial intelligence, Jason Lobel, CEO and co-founder of SwiftIQ, revealed to attendees at the 2018 NACS State of the Industry Summit near Chicago “how some of our peers and Amazon are making retail smarter using insight automation and artificial intelligence.”
“It’s a scary topic,” said Lobel. “Data, insights, analytics are very overwhelming. My goal for you when we leave is growing your business using data, building your EBITDA. It feels good.”
Lobel founded the convenience analytics platform in 2013 in response to the announcement of Amazon Fresh, the Seattle e-commerce company’s grocery delivery service.
Using data-driven automation and AI, SwiftIQ enables retailers and vendors to measure, share and enrich highly granular basket-level transactions and other retail insights. SwiftIQ facilitates retail analytics through high-scale automation and AI to optimize store-level execution, category management, promotions, merchandising and pricing. It analyzes more than $100 billion in point-of-sale (POS) data serving large food and beverage retailers and their suppliers.
The company works with about one-third of the top 20 convenience-store retailers, as well as with smaller c-store retailers in the 20-store range.
During the "Making Retail Smarter via Insight Automation and Machine Learning" session, Lobel broke the subject down for retailers and highlighted some examples of AI in action ...
Amazon is competing everywhere, Lobel said. It bought Whole Foods and launched Amazon Go. Its cost to enter a new market is almost nothing because it has the best-in-class data platform. And it can do so much with data—measure everything and create so much automation that it can be competitive and take costs out of the system.
That's a critical point, Lobel said.
Amazon uses a lot of automation and AI—from the supply chain to the store and customers. And it’s easy for them because they control all three of those channels. It’s hard for c-stores to do this because there are a lot of parties—manufacturers, wholesalers and others—in the ecosystem and they have to find a way to share data and work together better to replicate what Amazon is doing.
In 2002, Amazon CEO Jeff Bezos said he would fire anyone who did not have an application programing interface (API) so that he and his team could access their data sets. This was a response to when, 15 years ago, he realized that Amazon had to go into mobile commerce. “A lot of the innovation you’ve seen from Amazon all stems from the fact that they are amazing with their data and they can basically get at any single point into other applications and analytical systems,” Lobel said.
APIs are open to anybody. A developer can make money going on to Amazon’s website and building apps. Its most famous open system is Amazon Web Services (AWS).
APIs are key to unlocking raw data. Raw data comes in as Amazon sells something, and the APIs allow users to interact with that data through any digital front end. So Amazon is using them as dashboards to get business intelligence. Amazon takes it a step further, allowing customers can reuse them.
“Why would anyone want a button to put on their washing machine? I’m not going to press this and have Tide come to me when I’m out of detergent. That’s just silly,” said Lobel of the Amazon Dash button concept. But that point of view is missing the point, he said. With Dash, Amazon is testing a replenishment system that can be embedded into partner apps. This is the way of getting it out into the market to let people test it. Anyone can start using Dash in about 10 lines of code in about 10 minutes, he said. Amazon's Dash Replenishment Services (DRS) allows connected devices to leverage Amazon’s retail platform to build automatic reordering experiences for customers.
“It goes way beyond a button that you press,” said Lobel. “It’s really about how do I automatically enable replenishment?”
Cherry on top
What are some of c-store retailers’ peers doing in AI to create value? Lobel provided a concrete example. In 2009, Coca-Cola debuted Coke Cherry in its Freestyle fountain beverage dispensing system. Through the learnings of having so many people interact with the product to create data, they actually commercialized a packaged beverage product, Sprite Cherry, he said.
AI in the supply chain
Here are some areas to which people are applying AI and problems they are trying to solve, according to Lobel:
- Autonomous trucking. Examples of AI in the supply chain include Anheuser-Busch’s Otto self-driving beer deliveries. PepsiCo, Tesla, Walmart and Meijer are all ordering self-driving trucks to cut down on their supply-chain costs.
- Shelf management. “This is getting a lot more visibility in grocery now,” Lobel said. Walmart and regional grocers are deploying robots that scan three times faster than humans, and they’re more accurate. They look for out-of-stocks, misplaced labels and incorrect prices. “They’re about 50% more productive,” he said. There is now image-recognition technology that can go down the aisle and tell what is on the shelf. “I don’t really see this impacting convenience due to the small box size,” said Lobel.
- Pricing. Electronic shelf labels such as those Kroger is testing will lead to dynamic price optimization, Lobel said. Whole Foods is also using them.
- Personalization. Kraft Heinz is using AI to personalize billions of recipes. All of those recipes have a lot of content. That yields data including web behavior and email history.
- Mobile. There are techniques for matching mobile-device location data with receipt-level data, including product, brand and time stamp. “Receipt-level data is the new king,” said Lobel.
The building blocks of insight automation include:
- Measuring return on investment (ROI), including promotions, displays, coupons, etc.
- Smarter assortments based on revenue and value to store.
- Optimizing bundles to maximize success of promotions.
- Enriching data to optimize analysis and maximize insights.
- Reporting performance to achieve sales and market-share objectives.
- Tracking new items to analyze their performance.
AI “is not easy,” Lobel said, even for Amazon. Earlier this year, reports said that Whole Foods was experiencing shortages and out-of-stocks. “Big mistakes,” he said. “Automation in AI is very complex, and it requires a lot of testing, and it requires a long-term commitment. ... If you’re going to go down this path, which I really encourage you to do, just know it’s really hard, and even the best companies with the smartest engineers are constantly tweaking and tweaking and tweaking.”
Lobel praised the leadership of The Parker Cos. President and CEO Greg Parker, who said the Savannah, Ga.-based c-store chain is committing to investing in building AI-based applications. “I haven’t heard a lot of convenience retailers commit to that type of investment,” Lobel said.
But he said retailers are going to make mistakes. “You can’t get knocked off the horse and stay down,” he said.
Data scientists don’t want to work in retail, said Lobel. It’s tough for retail and consumer packaged goods (CPG) companies to find analytics and engineering talent. Most of the demand is in professional services and manufacturing. And one data scientist is limited in expertise. A whole team is required.