Big data analysis has the capacity to revolutionize the postal industry, and an increasing number of posts around the world are beginning to experiment with it.
Dr Jose Anson, an economist at the UPU, believes that the potential benefits of big data projects to postal organizations are so great that for those that have yet to dip their toes into the water, the question should not be ‘if’ but ‘when’ they start making plans to start exploiting it. “Posts should adopt a big data approach now,” Anson says. “Big data is a must to develop innovative products and services to respond to the needs of today’s customers.”
What is big data?
Big data projects aim to analyze the large quantities of data that organizations collect, to help solve business problems or provide insights into the way a business is operating. But despite the name, it involves more than just large amounts of data.
Back in 2012, information technology research group Gartner defined big data as being characterized by three Vs: high volume, high velocity and high variety. In other words, as well as large quantities of data, it involves data that is generated at a very high rate and comes from several different sources.
Many sectors of the global economy have already involved themselves in big data projects – notably the retail sector – but Anson believes that the postal industry is uniquely positioned to benefit.
The main reason is that there is so much data available about the basic operation of a postal service – moving items between two points. For example, the US Postal Service (USPS) scans mail and parcels up to 11 times before they are delivered, which represents a potential 1.7 trillion data points from this source alone.
Katrin Zeiler, a big data expert at DHL, agrees, “Big data and logistics are made for each other,” she says. “That’s because it is involved in many sectors, with many customers and many data sets. We have a lot of data and at the moment it is not being used in the optimal way.”
Big data, small cost
One reason that it is now viable for postal organizations to attempt to extract valuable insights from the data they collect is that the cost of storing and processing this data, and the tools required to analyze it, have fallen over the past few years.
Storage costs in particular have fallen sharply, and the emergence of software, such as the open source Hadoop project – an application synonymous with big data analysis – means that very large data sets can be processed and analyzed cheaply on low-cost commodity servers rather than requiring expensive high-end computers.
The economics of big data analysis has also been transformed by the emergence of cloud computing. That’s because it enables organizations to store their data at very low cost, and pay for the cloud computing resources they need to analyze it on an hourly basis instead of having to make huge up-front investments in server infrastructure that may not be used very frequently.
Benefits to posts
The potential benefits that big data analysis can bring to postal operators fall into several categories, including lower costs, higher operational efficiency, a better customer experience and the possibility of developing successful new business models and products.
One example of where big data might help postal organizations is route optimization, and UPS is one operator that is leading the way in this area. Since 2008, UPS has been using telematics technologies to gather data from several aspects of fleet operations. GPS tracking equipment and vehicle sensors are used to capture data related to engine monitoring, speed, mileage, number of stops, miles per gallon and safety. Data is gathered from more than 200 data points for more than 80,000 vehicles every day.
The data is analyzed to develop operational improvement strategies that also have environmental benefits. In fact, in 2012 UPS reported that as a result of its big data route optimization project it had eliminated 12.1 million miles of driving due to reduced distance per stop, and since 2001 had saved 39 million gallons of fuel through route optimization.
UPS is now working on the next stage of its big data route optimization project, with the launch of new software, which it hopes will optimize nearly all of the firm’s 55,000 routes by 2017 (see Inside Track, page 25).
Other areas where big data might help posts include efficient staff and vehicle scheduling by using data to predict specific future requirements, and reducing last-mile delivery failures by analyzing data about the likely location or status or the recipient. For example, if previous deliveries have been successful after 4:00pm but never before, this provides useful information to increase the likelihood of a successful last-mile delivery.
DHL is developing a tool to analyze correlations between things like weather, flu epidemics and Google trends to predict parcel volumes and determine staff and vehicle requirements. Results suggest its use results in a 15-20% increase in efficiency. And Australia Post is using predictive analytics to develop daily customer sales forecasts and predict profitability of products.
Left: Emile Naus, big data expert and partner at LCP Consulting
Big data in other industries
To get an example of the value that may be hidden in data repositories, Emile Naus, a big data expert and partner at LCP Consulting, cites the example of a large retailer he worked with recently. It had been collecting large quantities of data from different sources, but hadn’t attempted to analyze it.
For an initial outlay of less than £250,000 (US$430,000), the company built up a history of its business over the past three years using daily data about individual stores and products. This amounted to several terabytes of data in all – not a vast amount in data processing terms, but a substantial amount nonetheless. “This wasn’t a large initial investment, but it paid for itself very quickly in terms of the insights it provided,” says Naus.
These included a clear understanding about which products were profitable and which were not – an insight that was worth millions to the company. It also gave management a view of the amount of inventory that was in the supply chain at any given time – reducing it resulted in additional savings – and an understanding of which products sold in some stores but not in others.
Another example from the retail sector that could be even more relevant to the postal industry is provided by the experience of DM-Drogerie Markt, a Germany-based drugstore chain that employs over 36,000 people across Europe. The company implemented a big data analysis project to predict how many staff should be rostered for each shift – a process that was previously carried out by managers using their intuition for how many people would be needed in the following few days.
DM-Drogerie’s system takes account of historical revenue data, arrival times of new goods from distribution centers, and external data, such as public holidays, road diversions and weather data (as this can have a great effect on consumer behavior). It provides accurate projections of staff requirements and enables efficient shift planning weeks in advance.
“The biggest step is turning the data you have into insights that are of value,” says Naus. “Once you have the insights, then deciding on the action to take is easy.”
Left: Multiple scanning of parcels by USPS provides it with vast amounts of data to improve services
Where to start
For posts that have yet to start exploiting the big data they generate, a key question is how they should start. Is the data they already collect sufficient, or do they need to target certain types of data that should be collected and stored?
“Our experience is that, without fail, customers have more data than they realize,” says Naus. “The typical problem with that data is that they can’t see the wood for the trees, and don’t know where to start,” he adds.
Naus’s advice is to start small by choosing a particular problem to solve. “It could be a specific cost reduction or a customer service issue – anything. But focusing on what you want to achieve is key.”
This advice is echoed by Adam Houck, a senior managing consultant at IBM. “What we have seen to be effective is when organizations start with a very specific problem, such as why there is poor service in this area, why the power bill is higher than last year, or why the budget has overrun. The point is that you must start with a specific business problem. The top-down approach is often not the right way to go,” he explains.
But Anson warns that privacy is an important issue that has to be considered very carefully. “One of the big assets that posts have is public trust. Some people think that their privacy could be jeopardized by big data, although they may be prepared to give up a degree of privacy to receive better service.”
New business opportunities
Anson points out that posts have the opportunity to analyze and exploit their big data for purposes beyond their primary business goals. “It could be that the insights from your data enable you to start a new business operation that is not about delivery,” he says. “For example, data collected in real time from postal operations may be an excellent lead indicator of economic activity and consumer confidence. So the question is whether you want to share or sell those insights.”
DHL has even gone so far as to use its big data to create a product called DHL Geovista, which it sells to businesses. The product analyzes information about geographic locations, businesses in that area, buying behavior and social information to help small companies choose a place to set up business. “This is not a product designed for internal use – it is a new product we have developed to sell to the outside world,” concludes Zeiler.
Author: Paul Rubens
August 13, 2014
