I’ve been leading a few calls for the International Institute of Analytics, as a faculty lead for the manufacturing group.  We are preparing for the Chief Analytics Officer meeting coming up in Chicago in June.  One of the issues that comes up is where the analytics function should reside – in IT or in the business function?

We have heard from people on both sides.  The people in the business function complain that IT is “old school”, and that “we can’t rely on IT”.  They are saying “we need to down the analytics platform, and define the roles and responsibilities of that area.”

Another individual from a newly formed mfg. analytics group noted that a lot of what they had done was in constraint management and inventory and demand.  But most of the work that had been done was done internally – and not by IT.  IT “stores it in places, but doesn’t know how to create a good analytics foundation.”

I believe that IT organizations need to be viewed as a partner.    For instance, one IT analytics leader we spoke with noted that “we want to be partners with the business – and understand business needs and what drives the business.  We are focused on building an enterprise data warehouse, and providing the tools (in this case, in an Oracle environment) that allow people to click and drag and build their own reports, to get summary trends and detailed level data on what customers are ordering on a day by day basis for any product line at any region in the world.  It is viewed as “putting the platform out there and making it easy for people to create their own analytics.”

Another issue that comes up often is related to the process used to focus on an analytics problem and application.  This is an area where analytics can play an important role in leading the business function team to the right endpoint.  You need to be able to provide people with a “can-do” attitude.  This means being able to provide a definite process for engagement, which begins with identifying the problem, and building a set of hypotheses around what the relevant data are important that can provide insight into the problem.

One of the things that is important is being able to open up your ears and eyes to drive creative thinking into other types of data that might be available that are not currently in your standard database.  Is there proxy data that can be used to approximate other related variables that can provide good insights into the parameter of interest.   Are there other pieces of data that might be directly related but correlate to what we are trying to solve?  This could include public domain data or economic data.  Don’t just limit yourself to one or two data sources, but drive creating thinking.  This is what I mean by a process.

In the supply chain area, we are certainly seeing a strong need for analytics in areas like collaborative forecasting, inventory management, production planning, and logistics network design.  One of the big themes is the need to drive end to end integration – so that demand sensing algorithms can provide early warning of surges (or slowdowns) in demand, that can be tracked and driven back into the supply chain, beginning with improved warehouse and distribution management, transportation management, production planning, and supplier capacity and order collaboration.  People need to have metrics that drive transparency on events both in the short-term, but that also provide input into broader strategic decisions, such as longer-term needs for capacity, or even slowdowns that translate to postponing major capacity investments.

There is also a huge need for analytics in the area of talent management for supply chain and manufacturing.  We are working on a project looking at future talent requirements, and understanding what you need to be doing today that will translate into requirements for people and talent five to ten years from now.  We are working on building a talent management analytical tool that we believe is unique, and are looking for people to help validate it.


6 Responses

  1. Arun Gupta

    May 31, 2013 @ 5:05 pm

    I strongly believe that Business Analytics need to reside with the business with IT providing the technical and infrastructure expertise in facilitating business rapidly develop, validate, and embed outcome of these experiments into standard business processes. The key operative word here is “rapidly”; business needs to determine the level of accuracy and completeness acceptable to validate the model. Striving for 100% accurate and complete data will derail initiatives due to the time it takes to make this happen. Further, it is the business who owns and knows the data and needs to define the governance for data management, IT is just the holder of data. Now, if this was an Analytics initiative that was IT related instead of business, then IT needs to be the owner.

    What businesses really need are people who are IT savvy, capable of thinking IT and its multiple dimensions, but also have strong business acumen to be the bridge between IT and business. Further, Analytics initiatives need to be worked on in cross-functional teams including at least business, finance, and IT representations with a team member skilled in advanced analytics.

  2. Kevin Wurtz

    June 1, 2013 @ 1:48 pm

    I agree with your assessment that IT and business need to partner in developing a platform for mananging analytics. The conversation has to include what business needs to chart its course and what IT can truly deliver. You make a valid point in stating that we have to be open to various sources when gathering data. The caveat to state is that most searches don’t filter out but rather filter forward, and valuable data sources may be excluded. I am involved with several colleges in the area of operations and supply chain management as well as a not for profit that works with local manufacturers and would like to stay informed as to your efforts on developing the analytics for talent management.

  3. Chris Dyke

    June 1, 2013 @ 7:23 pm

    I agree that the first decision should be does the analytics belong in the value chain or in IT, however, I think another good question to bring up is where does the analytics section reside in the organization structure? Should analysts be fragmented into Sales, Marketing, Operations, and Supply Chain, or should they work cross in a cross functional structure? There are certainly pros and cons to both. We are structured where analysts are embedded within the functions. This allows for very customized support, however, it also results in much energy and effort to align across all the functions with the data being presented. Additionally, one analyst could be working on a product the other has already done.

    The bottom line with all of it is establishing a solid foundation in structuring the data collection and storage. Its important for functions to identify what to collect however I look for IT to develop the most efficient and user friendly way to store and access it. Disparate systems that the analyst has to tie everything together him/her self, result in additional time and effort that is wasted.

  4. Sandeep Srinivasan

    June 4, 2013 @ 5:14 pm

    Good post, Dr. Handfield. Two other things come to mind. Choosing the right key performance indicators is pertinent, so that company leadership is not making decisions based on skewed information. Second, (as companies gain wisdom/age) is to use big data tools to mine their backyard to (re)model indicators that will predict trends based on econ factors and unravel changing value propositions. However, the grunge work to maintain good data quality and keep these indicator models current will have to go on. Will be nice to know what insights come from the Chief Analytics Officer meeting in Chicago.

Continuing the Discussion

  1. Creating an Analytics Organization for Supply Chain — IIA

    September 17, 2013 @ 3:42 pm

    […] post originally appeared in Robert Handfield’s Supply Chain View from the […]

  2. Predictive Analytics to Reduce Manufacturing Risk | TIBCO Spotfire's Trends and Outliers

    May 24, 2014 @ 11:06 pm

    […] Additionally, manufacturers should be open to thinking about and identifying data other than what typically resides in a database, he notes in a separate post. […]

Some HTML is OK

or, reply to this post via trackback.