Elliot Grant, founder and Chief Technology Officer of HarvestMark, a fresh food traceability and insights platform, describes himself as someone “very passionate about the use of data to allow companies in and outside the produce industry to make better decisions.”
According to Elliot, “collecting data and the willingness to ask questions that challenge the status quo can sometimes lead to great decisions for your company.” Data doesn’t need to be big to be interesting, but big data, which is the most useful, cannot be managed with the tools available in a regular desktop. Such data sets are so large that analyses need to be broken up across multiple servers with tools like Hadoop and distributed virtual machines.
Big data was defined by IBM as consisting of four V’s. “The first one is volume, meaning there is a lot of data; the second one is velocity, because data is constantly coming at you very quickly; thirdly, it has variety, as data comes in many forms, including audio, video, texts, binary or machine code; finally, it has veracity, which means “quality”. It is all about making sense of a messy and unstructured amount of data,” explains Elliot.
In 1956, engineer Bill Fair and mathematician Earl Isaac came up with an algorithm to calculate credit risk that completely changed the way business was done. Fifty years later, we look at the arrival of Netflix, which also handles massive amounts of variables to offer recommendations to its customers. It is a learning algorithm that constantly improves with interaction.
When applied to the produce industry, this sort of variable management can be used, for example, to create mathematical algorithms able to predict what the product’s quality will be every day after shipping, taking into account variables such as temperature or packaging. “Once you have a mathematical model of your produce, you can measure the potential effect of changes in the supply chain,” explains Elliot.
For example, over a period of seven months, HarvestMark measured very large amounts of data on the quality, appearance and colour of bananas in order to optimise the ripening process, as the number of days a shopper gets after purchasing the fruit is a key indicator of satisfaction. The firm looked at the ripeness of thousands of bananas on the retail shelf and calculated the number of colour points the bananas ripened per day.
Analyses like this, taking into account factors such as the country of origin, type of ripening room, ripening methodology, packaging style, etc., enable companies to deliver bananas of the right colour that ripen at the required rate on the shelf. “As bananas are the number 1 produce category by volume, even small improvements like this can have a significant impact,” assures Elliot.
While the previous examples handle relatively few variables, other processes, like the scanning of products by hundreds of thousands of consumers with their phones, is a true example of big data. “People do this to get information about the origin, or to look for recipes, but through those scans we can calculate how old the product was and where the customer was located and find out, for example how fresh the product is on each U.S. state in each quarter of the year, or where there are gaps to be filled in the market,” affirms Elliot.
When comparing various products, there are all kinds of variables customers could care about, such as colour, sweetness, consistency, packaging, etc. “Our job is to measure every variable for each product. We could, for example, discover that in terms of sweetness, variety A is neck and neck with variety B, but half way through the growing period, B starts to drop off. As a grower I can then be confident that if I plant variety A, I will obtain a consistent sweetness throughout the season. It is all about correlating what customers want with the data we have access to,” explains Elliot.
The founder of HarvestMark explains that a very important step in big data management was taken with the implementation of standard identifying labels, currently used on around 4-6 billion cases of produce moving through the United States supply chain every year. Such tools offer great opportunities for analysis to improve efficiency, but “it is also great in terms of traceability, in case of problems, the source of the product can be easily identified.”
A very interesting way to personalise the use of variables is the ShopWell app, used by over a million people in the U.S. “You tell the app who you are and introduce information about yourself; for example, I’m trying to lose some weight, I need a heart-healthy diet, I have a specific allergy, etc. Then, in the grocery store, anything you scan will be rated as more or less suitable for you, using the algorithm to offer alternatives, just like Netflix does,” explains Elliot.
Each scan provides data on who the customers are, what store they are standing in, what products they are looking at and whether or not their recommendations were accepted, which provides very interesting feedback on how people shop, as well as the profile of customers for each specific store.
And then Elliott goes into his inside pocket and takes out: a Google Glass! Elliott believes that Google Glass will be used a lot in future by growers. "The grower can be on the field and take a photo with his glass from a tree and send it through." The grower can do a lot of communication or work through Google Glass while staying on the field.
“The future will definitely be all about mass personalisation,” concludes Elliot Grant.
For more information:
Elliott Grant
HarvestMark
Tel: +1 650.264.6200
sales@harvestmark.com
www.harvestmark.com