Practical Big Data series, part 3
In my previous blogs on Practical Big Data here and here, we searched for the best approach to big data that would best provide business value. We called this approach “Practical Big Data”.
Today we will be looking at real-life examples of Practical Big Data. The best known examples of big data in general tend to revolve around digital B2C services (Facebook, Twitter, Google, Amazon, eBay, etc.). However, being mindful that the reader might be interested in a B2B perspective as well, we also will give a few examples from this realm.
Now, let’s look at a few examples per each level of Practical Big Data:
1. Power up processes
Business process latency is a big waste of money in many enterprises. Companies around the world are spending billions of dollars in slow business processes such as revenue management or finance.
Imagine if instead of spending 27 hours to optimize the pricing of your entire assortment, you could do it in only one hour? This is what the US department store chain Macy’s accomplished. With the help of big data technology (combination of Hadoop and in-memory analytics), Macy’s is now able to re-price their assortment of ~73.000 SKU’s and react to changes on the market several times per day. This not only saves working time, but more importantly ensures optimized prices at all times both in the physical stores as well as Macy’s online channel. The benefit of an enhanced price optimization process is that a one-percentage-point improvement in average price of goods and services leads on average to an 8.7 percent increase in operating profits according to a McKinsey study.
In a second example, the construction industry products manufacturer Hilti has empowered their direct sales force with real-time analytics. Earlier it took Hilti roughly 3 hours to produce a customer report as the process requires scanning through all the 9 million customers and over 50 million transactions in the database. Now with big data technology from SAP (SAP HANA) this takes only 3 seconds – an improvement factor of 3.600x! Today Hilti’s direct sales reps can engage their customers in real-time with fresh reports wherever they travel. This leads to better interaction with customers and higher sales force efficiency.
In a further example, Deutsche Telekom achieved a 2.000x time improvement in part of their financial close process. Running a PoC against their former ERP system that took 35 minutes to run the process, with SAP HANA the time was reduced to mere 2 seconds. With this fast financial process, real-time finance and soft close whenever you want is not that far away!
2. Engage in real-time
Digital players use predictive algorithms with great success in engaging their customers. For example, LinkedIn’s feature “People You May Know” spawned from a big data initiative. PYMK has been accounted for providing 30% better conversions than Linkedin’s other ad campaigns and contributing to millions of new users for Linkedin.
Amazon and eBay are using similar algorithms for their NPTB (Next Product To Buy) recommendations. NPTB recommendations are calculated based on millions of user sessions and thousands of purchases made by users with similar profiles. These digital players are of course very secretive about the exact workings of their algorithms as they invest annually millions of dollars to fine-tune them.
For brick-and-mortar companies real-time predictive promotions may seem too far off – but it need not be so. The Japanese electronics retailer Yodobashi uses in-memory analytics (SAP HANA) to run their loyalty point calculation process in only 2 seconds. With 5 million loyalty cards, this used to take 3 days as a monthly batch process. With this significantly faster process, Yodobashi is able to make predictive personalized promotions to the shopper while they are still in the store. Where most retailers still only mine their loyalty card databases for targeted email campaigning, Yodobashi has already turned big data into practical, highly interactive, everyday use.
Also B2B companies can learn from the above B2C players when engaging their customers online. Tailored recommendations can be equally powerful if they are relevant and delivered to the end users in real-time. However, the data sets and predictive algorithms would in a B2B case appear somewhat different, since there would not typically be millions of user sessions and shopper baskets to analyze. Rather, in B2B e-commerce a rules-based predictive engine would be used, where the key inputs would be factors such as the end user role, customer’s contract specifics, omnichannel purchase behavior, product margins, and various predefined merchandizing and promotional decisions made by the product managers.
3. Transform the industry
Travel is a prime example of an industry that has been totally reshaped by digitalization. Not so many years ago to get a travel ticket you still needed to call your local travel agent and get the ticket in postal mail days later. Today, of course, everything is already digital and real-time.
For Amadeus, a company providing technology to the travel industry, this has meant a significant investment in big data and a vast change in their offering. Amadeus has created innovative products that utilize big data such as Extreme Search where the consumer only enters her overall budget, number of passengers, length of time for the trip, and the minimum temperature at the destination to receive trip proposals online. For Amadeus, transforming in pace with the industry and embracing big data early has been a matter of life and death.
Users of Netflix have already for years received recommendations on what movie to see next. Netflix uses big data and a smart predictive algorithm called Cinematch to predict the score that a particular user is likely to give to a movie. Netflix deemed this feature as such a critical success factor that it publicly offered a $1M prize to the team that could most improve their algorithm. Over the years 2006-2009 over 44.000 submissions were made by over 5.000 teams, and the winning team managed to improve the algorithm’s predictive power by 10% in 2009 hence receiving the million dollar prize. This has helped Netflix to become the leader in streaming video and to transform the entire movie distribution industry.
For many B2B companies the urgency to create data–driven offerings may not seem as immediate. Yet, some of the leading companies are already taking bold steps to transform their industry proactively:
One such company is John Deere that gathers sensor data from its tractors. This sensor data is analyzed to provide predictive maintenance i.e. to provide the user with predictions when a particular part of the tractor may be at risk of failing and should be repaired or replaced to avoid expensive downtime. Furthermore, John Deere also collects other data such as weather and soil data to provide their customers with recommendations on the best time to plant seeds, to harvest, etc. Information is pushed to the dealers and end users through various channels such as FarmSight and MyJohnDeere.com in both web and mobile versions.
In this way, rather than selling only tractors, John Deere has turned its unique selling proposition towards best-in-class productivity. Tractors are a commodity that can be bought from lower cost suppliers from all over the world. However, productivity – i.e. maximizing the farm output – has a totally different business case. I believe that there are many B2B companies who could benefit from John Deere’s thinking of how to harness big data to reshape their industry and to deliver practical, data-driven end products to customers.
To see slides on this Practical Big Data series parts 1-3 on Slideshare, click here!
