Sears Holdings is betting on its innovative rewards program to spur the company’s growth. It is a path that Walmart can’t match because it doesn’t have a loyalty program — its best hope of understanding customer identity is through customers who use its credit card. Sears has a very intensive big data program to drive customer loyalty; the sophistication surprised me and should interest investors.
I interviewed Dr. Phil Shelley, CTO at Sears Holdings Corporations about Sears’ use of the open-source Hadoop analytics platform kind of technology behind this award-winning program. Shelley was a keynote speaker at the Fusion 2012 CEO-CIO Symposium in Madison earlier this year.
Sears is doing amazing things with technology; it has a hit on its hand with its loyalty program. Retail trade magazines have focused on the loyalty program, the technology, and the significant investment which Sears has made as indicators of improved performance in the future.
In naming Sears Holdings the Master of Enterprise Loyalty (Global) COLLOQUY said the company recognized that it had great brand recognition including Sears, Kmart, Lands’ End, Craftsman, Kenmore and DieHard, but it lacked customer recognition.
“Their best customers didn’t know they were best customers… a risky situation in a love them or lose them retail environment…
“The solution was 3 years in the making, which included programming that would capture, analyze, and report on customer activity at an individual level, across all 4,000 locations. The ensuing information would enable Sears to provide the differential treatment and personalized attention their best customers deserved…
“Results: Sears achieved an active member base in the 8 digits, exceeding the projected 36 month membership target in 17 months. Member spend is now identifiable, enabling SHC to understand member purchase history, leading to more relevant and targeted communications and offers. And just as anticipated, members shop more often and spend more per transaction than non-members.”
In fact, participation in the loyalty program is up to 80 million members and counting, according to the Sears Holdings website.
Sears Holdings manages that huge amount of data with the latest technology; Shelley joined the company work work on the transition from legacy systems.
Shelley said that big data and Hadoop allow the company to keep all its customers’ history to the lowest level of detail — “every customer, every SKU, every store, every point of history for as long as you want to and do statistical analysis on it. With Hadoop there really isn’t any restriction.”
Sears has been a leader among non-Internet companies in making the most innovative uses of big data, he added.
“We are bleeding edge on a large scale, Some of the innovations are just amazing. Now you can put all your data in one place and achieve a single point of truth and use it at a granular level that was pretty much impossible before.”
The tools can be used across the company’s operations. Fraud detection was an early use case but big data is also used for supply chain optimization, personal pricing, promotions, tracking marketing campaigns and locating and pricing overstocked inventory for clearance.
“We are re-engineering an old legacy company to become a big data company.”
John Perrone, Sears VIP and Clubs Loyalty Marketing Manager, said the program isn’t just about building loyalty but driving business results, according to CustomerManagementIQ.com. The key, said Perrone, is more comprehensively evaluating the level of customer engagement.
“They’re giving up their email or their mobile phone number to interact with us, and as long as we, the brand and the loyalty program, reply back with relevant information and information in the channel that this customer prefers, it’s a very successful engagement and interaction with that customer.”
According to Perrone, maximizing multi-channel customer loyalty therefore banks on the value of the trade-off. If loyal customers are expected to convey worthwhile information on their consumption and advocate on behalf of the brand, they need to see a “return” on that loyalty.
To Shelley, getting the customer’s opt-in and delivering value in return are key to success.
“We [as people] want personalized experiences. They mean more to us and are more relevant. I hadn’t thought about it much until we were getting into that level of personalization because it wasn’t possible before these tools became available. Now you can get very personalized.”
Cimphoni’s Rick Davidson, co-chair at the conference, said many IT organizations are befuddled by big data, and Shelley agreed.
“Everywhere we interface with others, this is being led by business people and not by IT. IT managers are struggling to get their arms around the big data open-source movement around Hadoop when they have built on expensive proprietary data warehouses. The IT people are not ready for it; the skills are very, very different. I would guess only 20 to 30 percent of IT management at the meetings I attend have a decent knowledge of what Hadoop is.” Many can’t spell it, he added.
Luke Lonergan, CTO of Greenplum who also spoke at FUSION 2012 (See post below) speaks about Hadoop traps that companies fall into as they deploy big data solutions. Shelley said they are a result of a serious skills gap.
“The way you would load and manage data is profoundly different. It is very easy for almost any IT shop to stand up a Hadoop instance, even on the cloud. Getting a proof of concept is easy, but making a production scale operation and adding business value, that is where people fall over all the time. There are so many things to consider — data modeling, injections, constructing the data, how to write the query, how to consume the data and how to integrate with legacy systems. You have a whole host of questions. People do get befuddled and lost and you see them in a proof of concept for months or even years.”
The tools for using big data have been maturing, but only a few firms are ready for it. A major challenge is thinking of the questions to ask of the data.
“The imagination is more the limit than the technology, the first time in my career that has been true. People used to say they needed some information but it would take a week to run. Now that limit is gone; the limit is people’s imagination.”
In a letter to shareholders earlier this year, Sears Holding Chairman Eddie Lambert offered hope to shareholders who may have been disappointed with financial results.
“We have significantly grown our Shop Your Way Rewards program, improved our online and mobile platforms, and re-examined our overall technology infrastructure. We believe these investments are an important part of transforming Sears Holdings into a truly integrated retail company, focusing on customers first.”
Lambert further emphasized the importance of technology to the company’s success in announcing his choice to fill the CEO slot which had been an interim appointment for three years. He selected Louis J. (Lou) D’Ambrosio, a Harvard MBA with experience at IBM and Avaya.
“From the beginning of our CEO search, we were determined to find a leader with information and technology experience who could catalyze the transformation of our portfolio of businesses in the context of the evolution of the retail industry that is occurring more broadly,” Lampert wrote in his chairman’s letter that accompanied the release of fourth quarter results.”
This was no surprise to Davidson. He has said that companies will look for IT experience in a CEO just as they have looked for experience in sales or finance now.
“Going forward, I don’t think CEOs are going to be considered fully qualified unless they have some technology background, not necessarily application development, but something like running a large IT project.”
At the Fusion 2012 CEO-CIO Symposium, Shelley charmed the attendees by de-mystifying Hadoop with some help from the audience. Tearing a page from the program, he ripped some text-heavy content into thin strips which he asked participants to search and count the number of times a letter appeared.
Positioning Jody Franz of the Outlook Group as a traditional relational database, he pulled six men from the audience to be the servers in parallel processing and a seventh who acted as MapReduce and distributed the strips of paper and then collected the results in a fraction of the time it took Franz to count through her selection.
Hadoop, named after the toy elephant of the founder’s son, is open source, free and it will do practical work, Shelley told the audience.
“If you have ETL (Extract, Transform, Load — a way to move data from one system to another) jobs that run for hours, they cause problems because you have to wait for the ETL to finish before the data is usable.” Hadoop can run queries within seconds of ingesting data,and once it is installed, the data never has to move again. Heavy batch jobs, even in COBOL, that take 90 minutes to run on a mainframe can be done in Hadoop in two minutes. To read one terabyte, sort it, write it back and sort it again could take days, added Shelley. Yahoo did that last year across 4,000 nodes in 62 seconds.
Hadoop clusters are extremely robust since the data is duplicated across three nodes; if one fails,, work is directed to the other two and when the node recovers, the data is written back to it.