Everyone knows it’s not easy to get management support for Big Data projects. The ROI can be soft and the payoff long term. One way to get senior leadership buy-in is to frame the conversation in terms of returns today, not tomorrow. That’s what Rick Hassman, CIO of Pella Windows did, when he proposed a pilot project; he focused on immediate productivity gains.
Privately owned Pella designs, tests, manufactures, and installs windows and doors for new construction, remodeling, and replacement applications for home and commercial markets. Pella started thinking Big Data for three important reasons:
- It was quickly outgrowing its relatively new database structure as business people flooded its analytics group with requests for information.
- It was keen to tap into external sources of information like real estate sites and government.
- It wanted to leverage unstructured data generated through its call centers.
But big plans didn’t get management excited. Instead Hassman sold Big Data on the premise that it can increase productivity by helping people solve current business problems faster. Instead of a big, upfront investment, Hassman collaborated with a vendor and The Hackett Group on a four-week pilot that leveraged the cloud as an infrastructure. According to Chuck McMurray of Hackett’s Analytics Lab and Big Data team, the partnership lowered the project price tag and helped build a strong business case.
The project team chose a problem that the analytics team at Pella had been struggling with for over two years: identifying the features being requested by customers that required Pella to create so many customized quotes for its products. When salespeople created project quotes, options desired by customers were not always selectable in the quoting tool. Instead, while the system allowed salespeople to order the windows, they had to put instructions in an open notes field. The problem qualified for the pilot project on two levels. First, there was plenty of data. “We had millions of quotes saved, over 20-plus years of data,” Hassman said. Second, it included text and pictures.
Launching the project
It took two weeks to upload 15 terabytes of data to the cloud. “What we learned is that when you’re dealing with that much data, you have to ensure the host environment has the throughput and processing capacity to upload. But once the data was loaded, it took only hours to begin to make sense of it.
“When we got the analytics team what they needed, it took them half a day to look at the text and numerical fields together, identify the customer attributes, and start asking questions. The team was able to compare custom attributes to what customers were selecting.” This information was crucial as Pella’s product team decided what selections should be added to the quoting tool.
This was only the first stage of experimentation. With the quote data now uploaded to the Hadoop environment, it can be used in all sorts of ways. For example, in the future it can be combined with data about product quality to understand the correlation of customer and product information with any quality issues. The quoting data can also be combined with demographic information to better guide selling. Hassman added that “the old data warehouse methodology didn’t allow us to get to that level of detail.”
What’s ahead for Pella?
“Once we changed the way we characterized the technology, making it about shortening the process and improving productivity as opposed to gaining some future capability, enthusiasm took hold,” Hassman said. With this win, “we’re moving in the right direction for the business and the analytics team,” Hassman said. The natural evolution is to direct attention to other discrete data pools and begin to ask more advanced questions. By optimizing the efficiency of the analytics team and the business, Big Data and analytics can help free up the analysts’ time to begin asking new, more probative and advanced questions. “We will be able to ask questions we never could ask before.”
Learn more about Data – The Hidden Treasure Inside Your Business.