#Jang Saetbyeol
2017.11.23
Chapter 3. Does Big Data Ship Products?
Introducing the columns written by member of Amorepacific Group
ColumnistJang Saetbyeol
Amorepacific Amundsen Camp
Introduction
Hello. I'm Jang Saetbyeol. We are in the process of exploring a number of different big data practices along with our business flow. We discussed product planning when a product is "born" in chapter 1, and best big data practices in production and quality control in chapter 2. Now in chapter 3, we will explore big data practices in distribution and logistics, in which products produced in a plant are delivered to customers.
Does data place products?
The following figure shows a case of improving the layout plan for an open-air storage yard where products are loaded based on data. The left layout shows the previous state of the yard where products were arranged based on the experience of skilled forklift drivers.
-
Placement of products in open-air storage yard utilizing big data (Source : CLO)
This company collected the traffic and release data of the entire yard and made groups of similar products and analyzed what product groups are usually released together. It changed the layout of products based on this and as a result, it could reduce the operating distance of forklifts by 30%. This seems to be a small and simple case, but could be an example of small learning from big data analytics in distribution and logistics in that the company has gradually improved what it would do based on experience and on big data.
Quickly and accurately
Affordable price, fast turnover, trendy design...these are what represent "Fast Fashion", which has led the industry for some time.
Whereas the existing clothing companies planned and produced products four times a year according to the flow of seasons, what differentiates fast fashion is that they plan and release products faster at one to two week intervals in response to the latest trend and consumer reactions. Because of these characteristics, it became much more important to understand market trends and predict sales and demand, compared to the past when they counted on the sense of the designers.
ZARA, a Spanish clothing retailer, in partnership with MIT of the U.S., developed an inventory distribution optimization system that collects and analyzes the sales and inventory data of its entire markets around the world to achieve the highest possible sales. This system predicts sales for the next week by considering the sales and inventory of each store and the volume of products requested, past sales records and store display, and reflects the results to production and distribution planning. Based on such prediction and with an optimization algorithm, ZARA ships products directly from its two distribution warehouses in Spain to over 2,200 stores in over 90 countries around the world. This system reduces inventory, which is a crucial cost factor in the distribution industry, to zero, and optimizes inventory distribution to maximize the sum of sales of every store.
Whereas the existing clothing companies planned and produced products four times a year according to the flow of seasons, what differentiates fast fashion is that they plan and release products faster at one to two week intervals in response to the latest trend and consumer reactions. Because of these characteristics, it became much more important to understand market trends and predict sales and demand, compared to the past when they counted on the sense of the designers.
ZARA, a Spanish clothing retailer, in partnership with MIT of the U.S., developed an inventory distribution optimization system that collects and analyzes the sales and inventory data of its entire markets around the world to achieve the highest possible sales. This system predicts sales for the next week by considering the sales and inventory of each store and the volume of products requested, past sales records and store display, and reflects the results to production and distribution planning. Based on such prediction and with an optimization algorithm, ZARA ships products directly from its two distribution warehouses in Spain to over 2,200 stores in over 90 countries around the world. This system reduces inventory, which is a crucial cost factor in the distribution industry, to zero, and optimizes inventory distribution to maximize the sum of sales of every store.
-
ZARA moving ahead with its big data-based supply chain (Source : Wikipedia)
Currently ZARA is not as hot as before due to the advent of Ultra Fast Fashion, which features a production/distribution cycle of less than one week, but it was big data that supported the success of Zara, which once had the world's fastest supply chain.
Umbrella seller and shoe seller
Let's take a break with an old fairy tale. Once upon a time, there was a mother who had two sons: an umbrella seller and a shoe seller. On rainy days, she worried about her son selling shoes, but on fine days, she worried about her umbrella selling son, so she had to live with worries every day.
When I listened to this story in the past, I would wonder why she had to worry because at least one of her two sons would sell a lot regardless of fine or rainy weather. If she lived in the world of big data like today, she could be rid of her worries. She could predict which product will sell better on the day and have her two sons go to the market with the same product.
Recently, the home shopping industry is utilizing various technologies that are leading the Fourth Industrial Revolution. By analyzing data, they verify the impact of weather on sales and also draw optimal ways to assign their shows with seasonal products. In particular, insights obtained based on these analyses are immediately utilized throughout the business.
"G" home shopping channel in the domestic market analyzed air conditioner sales and weather data for the past 10 years and concluded that the number of days of heat wave for the previous year influenced sales for the first half of the current year. In the last summer (2016), which, as you know, was record-hot, there were a total of 24 days of heat wave with daytime temperature highs of more than 33 ℃, three times more than in 2015. Based on this result, the company started its air conditioner program from April, which usually starts from June, and posted a sales volume of 145% over the target in April and May this year. As the home shopping retailer experienced such success based on data, recently it decided to make investments for the future by constructing a cutting-edge complex distribution center. It is expected that the home shopping retail industry will continue to evolve.
When I listened to this story in the past, I would wonder why she had to worry because at least one of her two sons would sell a lot regardless of fine or rainy weather. If she lived in the world of big data like today, she could be rid of her worries. She could predict which product will sell better on the day and have her two sons go to the market with the same product.
Recently, the home shopping industry is utilizing various technologies that are leading the Fourth Industrial Revolution. By analyzing data, they verify the impact of weather on sales and also draw optimal ways to assign their shows with seasonal products. In particular, insights obtained based on these analyses are immediately utilized throughout the business.
"G" home shopping channel in the domestic market analyzed air conditioner sales and weather data for the past 10 years and concluded that the number of days of heat wave for the previous year influenced sales for the first half of the current year. In the last summer (2016), which, as you know, was record-hot, there were a total of 24 days of heat wave with daytime temperature highs of more than 33 ℃, three times more than in 2015. Based on this result, the company started its air conditioner program from April, which usually starts from June, and posted a sales volume of 145% over the target in April and May this year. As the home shopping retailer experienced such success based on data, recently it decided to make investments for the future by constructing a cutting-edge complex distribution center. It is expected that the home shopping retail industry will continue to evolve.
Joy to the world, my shipment is come!
One of the common concerns of online retailers and distributors is shipping. The ability to deliver a mass amounts of products quickly and accurately, leads directly to customer satisfaction and better corporate reputation.
Amazon, an online distribution giant, announced its introduction of anticipatory shipping previously in 2014. Anticipatory shipping means literally to ship products before a purchase is made by predicting what the customer will buy. The aim is to minimize the time taken to deliver the product to the customer's door. They ship products to their distribution base adjacent to the customer's location before he or she actually purchases them and ship them to the customer's door as soon as the order is taken.
Amazon, an online distribution giant, announced its introduction of anticipatory shipping previously in 2014. Anticipatory shipping means literally to ship products before a purchase is made by predicting what the customer will buy. The aim is to minimize the time taken to deliver the product to the customer's door. They ship products to their distribution base adjacent to the customer's location before he or she actually purchases them and ship them to the customer's door as soon as the order is taken.
-
Amazon's anticipatory shipping service (Source : Google)
For this, Amazon used massive amounts of customer data to find out the purchase cycles of specific product groups. The easiest target was daily necessities, which are necessary and re-purchased frequently. The company identifies patterns of purchase of daily necessities such as water, paper and detergents, and predicts "when" the customer will re-purchase "what" product.
Later, Amazon also introduced a more accurate prediction method that uses various online behavioral data rather than simply finding out the purchase cycle. Amazon finds out what keywords are used when the customer searched, what products they put in their wishlist, what products they searched for recently, and how long they stayed in specific pages or photos, and based on the results, the company ships some products likely to be purchased to a distribution center close to the customer's main shipping address even before the purchase is made.
Later, Amazon also introduced a more accurate prediction method that uses various online behavioral data rather than simply finding out the purchase cycle. Amazon finds out what keywords are used when the customer searched, what products they put in their wishlist, what products they searched for recently, and how long they stayed in specific pages or photos, and based on the results, the company ships some products likely to be purchased to a distribution center close to the customer's main shipping address even before the purchase is made.
I have already prepared it because I knew you would like it
Alibaba, a newly emerging e-commerce giant, also says the key to their success is their fast and accurate shipping service. Ali Express, Alibaba's online shopping mall, also predicts demand by region and ships products to the region in advance, and in this way, they are overcoming the challenge of such a broad territory. As they prepare their inventory through accurate prediction based on data, they can ship products to the door within a day in metropolitan cities such as Shanghai and Beijing. If customers order a product in the afternoon and receive it the next morning, and experience this again and again, this becomes a new level of reputation boost that can be only achieved through the customer experience.
Having tracked shopping patterns and demand for many years by utilizing its unrivaled amounts of data on the Chinese domestic market, Alibaba confidently says that it will improve the accuracy of its prediction further by reflecting data on income classes and product preferences by country to its prediction for the global market. In particular, China saw over 30 billion shipping service transactions during 2016, and online shopping accounts for about 70% of the total. As Jack Ma, Chairman of Alibaba, said the company would realize daily shipping transactions of one billion in six to seven years, it is interesting to see how fast the company will grow based on data and technologies.
Having tracked shopping patterns and demand for many years by utilizing its unrivaled amounts of data on the Chinese domestic market, Alibaba confidently says that it will improve the accuracy of its prediction further by reflecting data on income classes and product preferences by country to its prediction for the global market. In particular, China saw over 30 billion shipping service transactions during 2016, and online shopping accounts for about 70% of the total. As Jack Ma, Chairman of Alibaba, said the company would realize daily shipping transactions of one billion in six to seven years, it is interesting to see how fast the company will grow based on data and technologies.
Streamlining logistics reduces time and cost
-
Logistics system based on IoT and big data (Source : DHL)
DHL, the NO.1 global total logistics provider from Germany, foresees that streamlining logistics by utilizing big data will determine the competitiveness of the future logistics industry. DHL has actually integrated and analyzed various data, ranging from shipping history to weather, Google search keywords and online shopping behaviors. DHL explains that based on such analyses, it can raise the accuracy of prediction on the region, time and volume of demand, thereby elevating the efficiency of cost input, covering vehicle or ship/airplane placement or operation cycles.
DHL published , a report on big data trends in logistics, some years ago, and has taken the initiatives in promoting the importance of data.
In addition to cost-effective shipping, a fast response to logistics troubles is also crucial. In particular, mid- to long-term logistics, such as shipping of producer goods or inter-company shipping, count more on marine transport than short-term and consumer shipping, such as online shopping, and thus they are subject to various logistics troubles.
DHL published , a report on big data trends in logistics, some years ago, and has taken the initiatives in promoting the importance of data.
In addition to cost-effective shipping, a fast response to logistics troubles is also crucial. In particular, mid- to long-term logistics, such as shipping of producer goods or inter-company shipping, count more on marine transport than short-term and consumer shipping, such as online shopping, and thus they are subject to various logistics troubles.
-
Cello, Logistics solution based on data analytics (Source : Google)
"S" company, which offers IT-based logistics services, provides logistics solutions designed to predict and detect such risks based on IoT and data analytics, and suggests optimal measures. Data, previously recorded and collected by people at the logistics sites, is not collected and is instead sent automatically through sensors, and empty spaces in warehouses or harbors can be detected through sensors to predict acceptable cargo volumes and assign locations. In addition, the company differentiates its services by drawing optimal routes based on its big data analysis algorithm keeping various factors in consideration, including chance of delay due to natural disasters such as typhoons, and inevitable land logistics in departure and arrival locations.
As avoiding logistics troubles is crucial to not only logistics service providers but also their customers (consigner), big data-based approaches in this area are expected to spread throughout the market.
As avoiding logistics troubles is crucial to not only logistics service providers but also their customers (consigner), big data-based approaches in this area are expected to spread throughout the market.
Power of abnormality
One of the many assumptions of data analytics based on statistics is "Past events and patterns will repeat in the future." Based on this assumption, we predict the future or estimate what we don't know within a certain range of probability by using past or known data. However, what if there is any outlier in our data?
Although there are lotto winners every week, we are not convinced that we will win the lotto by any chance this weekend. This is because we already know that the event has an extremely low probability. Likewise, any outlier that deviates from a normal range prevents you from obtaining reasonable results from data analysis, and thus it needs to be examined and processed. (There are various technical methods to process outliers, such as exclusion and processing but in this article, we won't discuss this in detail.)
In logistics, managing such outliers is also crucial. For example, if the cargo volume suddenly rises or shipping time increases sharply, based on which standard should we judge whether it is a problem or not and establish countermeasures?
Even when the volume rises two or three times over the average, this could be just as high as the average cargo volume on Mondays or could be in a normal range during certain event periods, such as national holidays, Christmas or sale periods. If the effect of day of the week repeats on a weekly basis, it could be understood within some months, but what if there is also a seasonal factor? Or, what about the kimchi-making season when large amounts of big and heavy products are shipped, rather than a change in the number of products shipped? Although it occurs periodically every year, we might need data for many years to make a judgment based on non-human data.
This is why continuous data accumulation and quality control are necessary in the data analysis of any area, and is not limited to distribution or logistics. If you are assigned to conduct big data analysis at work, I hope you remember the principle "Garbage In, Garbage Out", which originates from computer science, but is also widely used in analytics.
Although there are lotto winners every week, we are not convinced that we will win the lotto by any chance this weekend. This is because we already know that the event has an extremely low probability. Likewise, any outlier that deviates from a normal range prevents you from obtaining reasonable results from data analysis, and thus it needs to be examined and processed. (There are various technical methods to process outliers, such as exclusion and processing but in this article, we won't discuss this in detail.)
In logistics, managing such outliers is also crucial. For example, if the cargo volume suddenly rises or shipping time increases sharply, based on which standard should we judge whether it is a problem or not and establish countermeasures?
Even when the volume rises two or three times over the average, this could be just as high as the average cargo volume on Mondays or could be in a normal range during certain event periods, such as national holidays, Christmas or sale periods. If the effect of day of the week repeats on a weekly basis, it could be understood within some months, but what if there is also a seasonal factor? Or, what about the kimchi-making season when large amounts of big and heavy products are shipped, rather than a change in the number of products shipped? Although it occurs periodically every year, we might need data for many years to make a judgment based on non-human data.
This is why continuous data accumulation and quality control are necessary in the data analysis of any area, and is not limited to distribution or logistics. If you are assigned to conduct big data analysis at work, I hope you remember the principle "Garbage In, Garbage Out", which originates from computer science, but is also widely used in analytics.
In conclusion
Today we discussed various big data practices in not only simple logistics that deliver products to customers, but also distribution, which connects producers and consumers effectively. We learned that decisions are made based on data in logistics and distribution, just as in the areas we explored in the previous chapters, but it is noted that big data in this industry are utilized in many different areas, such as customer-facing service areas (anticipatory shipping, ZARA's fast fashion, broadcast programming, etc.) and internal streamlining by companies (streamlining logistics costs, route optimization, trouble prediction, etc.).
I would like to wrap up today's column with the passage that impressed me the most when I explored big data practices in distribution and logistics.
"The way you can defeat 'fast' is not 'faster' but 'in advance'.
Also, I think we not only need to prepare products in advance and deliver to customers "a little faster," but we also need to prepare what customers want before they actually want it, as we discussed in the first chapter of this column. In the next column, I will come back with more interesting big data practices in connection with our business.
I would like to wrap up today's column with the passage that impressed me the most when I explored big data practices in distribution and logistics.
"The way you can defeat 'fast' is not 'faster' but 'in advance'.
Also, I think we not only need to prepare products in advance and deliver to customers "a little faster," but we also need to prepare what customers want before they actually want it, as we discussed in the first chapter of this column. In the next column, I will come back with more interesting big data practices in connection with our business.
-
Like
0 -
Recommend
0 -
Thumbs up
0 -
Supporting
0 -
Want follow-up article
0