In this special guest feature, Kevin Payne, Vice President of Marketing for Zest Labs, believes that by appropriately utilizing AI, machine learning and predictive analytics to know the actual remaining shelf-life of produce, grocers can more accurately plan for when and where to send it. As a result, there are fewer “surprises” because guess work (often based on old-school and inaccurate visual inspections) are taken out of the equation. This smooths out the bumps in inventory management and improves supply chain visibility. Born and raised in Silicon Valley, Kevin’s been a passionate and energetic marketer in high-tech for 30 years. He has successfully led marketing teams at startup and mid-sized companies including [24]7, Intelleflex, Evault, and NEC, has been a guest lecturer at several universities. Kevin holds a Bachelor of Arts in Marketing Management from San Jose State University and an MBA from Golden Gate University.
The fresh food supply chain, which links growers and suppliers to retailers, is facing unprecedented challenges with processing, transport and handling of product, among other complexities. The fresh food industry has traditionally been a fairly stable one but is now facing a dynamic competitive environment, changes in customer buying preferences, and increased costs that put pressures on already notoriously thin margins.
Statistics from a recent Progressive Grocer article Grocers Embrace AI to Optimize Supply Chain are cause for alarm:
- 43 percent of grocers are challenged by lack of real-time inventory visibility
- 48 percent of grocers rate their forecasting technology from very poor to average
- 43 percent of retail supply chain professionals say their technology can’t keep up with business demands
These statistics represent significant problems within the fresh food supply chain – problems that lead to food waste, dissatisfied customers and reduced product margins.
Without visibility into their supply chain, retailers tend to enter a cycle in which they run out of or low on products due to spoilage and waste, then over correct by buying too much which then also spoils because it doesn’t sell fast enough, creating a vicious and costly repeating cycle.
Much of this problem can be traced back to a lack of insight into the cause of waste and considering waste an accepted cost of doing business that couldn’t be fixed.
In all, because retailers don’t know the remaining freshness or shelf-life of the produce they receive, 12 percent of fruits and 10 percent of vegetables are wasted at the retail level. What other industry tolerates such a high level of waste?
Applying Artificial Intelligence to the Supply Chain
In what has traditionally been a comparatively low-tech industry, suppliers and retailers are increasingly turning to technology to address fresh food supply chain challenges, reduce waste and improve inventory management. Yet, at this point, only about a third of grocers are using AI in their supply chains and about one-in-four are moving toward that.
Retailers believe AI’s strongest potential to improve supply chain management is with the quality and speed of planning insights. It’s true. When it comes to improving the management and visibility of the fresh food supply chain, AI – when combined with the right data – can optimize the reduction of waste associated with overstocking and understocking, improve delivered freshness of the product and streamline inventory management.
Freshness Management is Key
But AI is merely a tool, albeit a very powerful one. What’s important is understanding how to apply it with a clear objective in mind. The insights gathered from an AI-based system are only as good as the data that goes into the system to help build and tune the models. If you start with the wrong assumptions, AI-based systems, no matter how elegant, will not provide useful results.
When it comes to reducing food waste, the fresh food industry has assumed that all produce harvested on the same date will have the same amount of freshness and shelf-life.
This is simply not true. The actual remaining freshness can vary by a number of days due to the impact of harvest conditions, handling and processing. If you assume that all pallets are identical, you will continue experiencing issues with delivered freshness and you won’t solve the problem. Simply put, that assumption will result in a “garbage in, garbage out” result.
Instead, you need to start by determining the freshness capacity (or maximum shelf-life) of the produce. This establishes a baseline. For example, you may determine that a particular lot of strawberries has a maximum freshness capacity of 12 days from time of harvest in ideal conditions. For strawberries, ideal conditions mean that they need to be quickly cooled to 34° F.
Then you need to collect data about the condition and handling of the produce at the pallet level from the time it’s harvested. By processing that pallet-level data (best collected with autonomous IoT sensors throughout the supply chain) with an AI-based machine learning system and predictive analytics, we can accurately predict how long that produce will last. Because the system utilizes machine learning, with more and more data gathered from more and more pallets, the system gets smarter and increasingly more accurate with its predictions.
This predictive model reduces waste and spikes in inventory because each pallet can be delivered to the retailer with sufficient freshness for sell through.
Use IoT Devices to Improve the Data
For an AI-based machine learning system to excel, it needs a sufficient amount of quality data. In the fresh food industry, access to reliable and accurate data has long been a problem. Traditionally paper-based, data records are often limited and error prone. For example, to monitor for cold chain integrity, such as making sure the cooling units on trucks are operating, the shipper would often insert one USB-based data logger in the trailer. Not counting the fact that the ambient temperature of the trailer has little to do with the temperature of the pallets of produce inside, the logger needed to be collected by someone, plugged into a computer, have the data downloaded and extracted into some other record keeping device. It’s slow, cumbersome and presents multiple opportunities to corrupt the data, if it’s even collected at all.
IoT (Internet of Things) condition sensors solve this problem. They can collect data – and lots of it – without manual intervention. By placing an autonomous IoT sensor in each pallet at harvest, for example, we can automatically collect data about that pallet every minute or at whatever interval we choose. That’s a lot of data to feed into our AI-based system that requires no manual intervention, so the chance of error is essentially eliminated and labor costs are reduced.
It’s AI-based Freshness Management
This freshness management methodology provides grocers with the ability to know the actual remaining shelf-life of a product they’re receiving at their distribution centers so they can improve planning and inventory management that reduces waste and improves operations and profitability.
By appropriately utilizing AI, machine learning and predictive analytics to know the actual remaining shelf-life of produce, grocers can more accurately plan for when and where to send it. As a result, there are fewer “surprises” because guess work (often based on old-school and inaccurate visual inspections) are taken out of the equation. This smooths out the bumps in inventory management and improves supply chain visibility.
Sign up for the free insideBIGDATA newsletter.
Speak Your Mind