Big Data? What is the Opportunity?

At a recent Manufacturing Leadership Forum hosted by Northwest Analytics, Dr. Lloyd Colegrove from the Dow Analytical Tech Center presented on Dow’s efforts to tame its data issue. He began by stating that suppliers don't understand the problem Dow faces. In an effort to convey the problem to management, the analytics group created a video depicting growing amounts of data and knowledgeable people leaving. This scenario is leading to repeated mistakes.

Until recently, plant support was reactive in nature, but proactivity is the goal. He noted that quality isn't the right focus. Once the lab tests are evaluated, it's already too late. Instead, Operations needs to utilize tools that allow them to make corrections before problems manifest in the product. Many times, the people are looking at the data are actually the problem. Context is key. Users must first ask, “Is the data any good?”

According to Dr. Colegrove, companies should have a goal to get their data and information working for them. One of the keys to this is the data must be live. By looking at data differently, it is possible to find new information. One of the examples he gave was to add context through adding new dimensions to the data. In one such case, Dow was being sued by a customer for providing bad product. There was no indication from the lab data that the product delivered was off-spec. When the data was analyzed by adding third dimension to the data, it became clear that there were four distinct products being manufactured and Dow’s customer only wanted one of them. The speaker made clear that multivariate views are not the same as a bunch of uni-variate views, however.

To achieve success, Dr. Colegrove believes that everyone involved must have access to the same tools. He expressed the importance of getting people involved in the process. Plants need to know the ‘Why.’ He also stressed the importance of automated actionable analytics. Most of the examples cited, only reviewed a small fraction of the total data signals. This data was initially evaluated for it importance and the ability for personnel to affect a change to the product outcome. At Dow, the analytic installations have been on a trial basis. To date, no site has requested the software be removed.

In closing, Dr. Colegrove expressed Dow’s analytic implementation as part of a movement from data to information to knowledge and ultimately to wisdom. This just won't happen; it requires effort and commitment on the part of users. He also recommended a book entitled, "Competing on Analytics." Dow’s journey into the use of analytics started with one plant and one man’s vision. To date, Dow has implementation in about 20 facilities; only +500 to go.

My take:

Dow doesn't express the benefits in actual numbers, but Dr. Colegrove did mention that the payout was in the millions. Much of what Dow is doing to achieve breakthrough results is based on techniques that have been around for decades (nearly a century, in fact). When looking to benefit from analytics, fancy analytics gets a lot of press, but small and simple is the way to start. One aspect of the tool Dow uses that can make the difference is the ability to pull together data from different data sources and glean insights in real time. This allows users more time to evaluate solutions and effect corrections. One other aspect is the user controls/adjusts the tool, not another department, which facilitates continuous improvement. This highlights the need for tools that make sense from a lifecycle standpoint.