Back to the Future of Analytics

Since we published more humorous stuff earlier today given the date, we thought we'd put a little more "meat on the bone." And what's better meat than some fine-aged beef?

The below article called "The Power to ask What If?" was published 9 years go to the day in Purchasing Magazine (which perhaps Questie is currently reading? Rest his soul).  It was an interview that Anne Millen Porter had with me regarding analytics -  and it's as relevant today as it was then.

Unfortunately, I still haven't seen the innovation in analytics that I would've hoped for back then.  It's a big and complex space. Here is a supply analytics paper in Supply Chain Management Review I did last summer and also a more detailed look at managing Procurement Analytics that we wrote in our multi-piece Procurement Information Architecture research series. Also, see more here from Spend Matters and Spend Matters PRO:

The future is unbounded for analytics. All it requires is some imagination, a few pedabytes of decent quality data, and some demonstrated use cases that will deliver targeted value to progressive organizations who've moved beyond forensic purchase history reporting. Whether you're a provider or a practitioner, you should definitely be thinking about how to raise your game in this area. And as always, contact us if you want to discuss what's next for you.

Here's the interview:

Businesses have used optimization or operations research - a proven, mathematics-based approach for complex decision-making - for many years. Airlines use it in flight planning. Manufacturers use it in factory scheduling. Transportation companies use it in advanced logistics management. But only recently - say within the past three to five years - have the pieces started falling into place to make optimization analytics an option for solving complex sourcing and procurement problems as well. That is, only recently have companies begun to achieve the levels of IT systems integration, data availability, data hygiene, professional skill sets in sourcing, and cross-functional cooperation that make it possible to pursue mathematically rigorous answers to complex, strategic sourcing questions like:

  • How can we evaluate a set of widely dissimilar supplier bids across multiple line items to yield the best number of suppliers (by line item) and the correct allocation of business among those suppliers (again, by line item) while, all the time, taking into account our profitability goals, supply risks, risk tolerances, corporate governance constraints, and broad array of internal customer requirements?
  • How can we weigh suppliers' bid prices against nonprice factors like contract terms, quoted delivery speeds, past performance, etc.?
  • What is the best supply-chain design - number and location of suppliers, distribution of work (make vs. buy), distribution and location of inventory, and flow of materials for this product we are selling?

PURCHASING recently interviewed Pierre Mitchell, vice president of research with Boston, Mass.-based AMR Research on what sourcing leaders need to know today about optimization and decision support technology.

Q: What are the major sourcing problems or pain points to which optimization technology is being applied?

A: Optimization has been around for decades and is standard fare in many supply chain processes such as production planning and scheduling or in logistics. In procurement, the most prevalent application area is bid optimization. Quite often, bidding practices involve saying, 'Never allow suppliers to influence your requirements to their favor, and then qualify suppliers on all non-price factors, so you can use price compression techniques during online bidding.' But that approach requires a lot of guesswork about supply markets and often puts suppliers in a position of bidding on business they can't deliver cost effectively. The business problem is that, for certain complex spending categories, you want to allow suppliers to always put their best offers forward based on what they know about their own markets, costs and capacities. You learn from suppliers by allowing them to influence your requirements based on what they know about their own costs - even if you don't incorporate all their ideas. You want to be a low-cost customer and construct the largest possible market basket and leave as many of the variables - like lead times and contract terms - as open as possible. This allows you to let suppliers optimize their operations so that they can pass on those savings to the buyer.

Sourcing of logistics services is a great example of where companies are using bid optimization successfully. You could put together a packaged bid, say five hundred lanes, winner takes all, but you might be forcing the service providers to bid on business that isn't price-attractive to them. They might need to subcontract in, say, fifty of the five hundred lanes; and then they pass those higher costs on to you. Bid optimization technology allows you to assemble large market baskets of requirements, then let suppliers cherry pick the pieces of business that optimize their cost structures. It allows them to offer you their lowest possible pricing.

Supply and price risk management are other business problems where optimization will be applied. Everybody's got some horror story, whether it's Ford, Palm, Motorola or whoever. Analytics let you look at your supply base, identify price, supply and other risks, and understand where you need to be hedging or using other risk management strategies.

Another big area for optimization is supply network design. It's where you answer questions like, 'Where should value be added? Where do we position global supply? Where should we attack leadtimes? What is the true Total Cost of Ownership and how do I minimize the tradeoff between TCO (including inventory), risk and customer satisfaction? You start to look at supply as a combination of inventory, plus suppliers' capacity, plus the capacity that suppliers could build. What this allows you to do is to not just fine-tune your internal network, but then to simulate the impact on your extended supply network if you are considering various business strategies to pursue. You ask, 'How can we reposition inventory in the supply chain to improve service levels and lower costs? What are the cost, lead time and inventory tradeoffs if we decide to source in a low-cost labor country like China? How will outsourcing decisions affect our responsiveness to customers?' You answer questions about how to engineer and reengineer the supply network so you can optimize your responsiveness to customers. Some leading edge companies are even using customer analytics like conjoint analysis and then are using their deep knowledge of their supply networks to conduct simulations and supply chain scenario planning at the time products are being conceived. They are basically assessing whether they can go after these new customer requirements profitably. Profitable growth is the single most important CEO objective that we see right now and supply management is what's going to increasingly separate the leaders from the laggards.

Q: So who's doing this already?

A: Really, it's companies that have their supply networks pretty much under control and they're looking to move to the next level. They've done all the obvious things in strategic sourcing. Now they're using advanced analytics to simulate future supply chains or supply networks. They're attacking their most complex spending categories in direct materials and also in transportation. And they're getting ahead of the game by trying to anticipate the supply chain impacts of moves they are planning to make in the future. For example, will such and such a move create a spike for milling requirements in Thailand? Can our supply network handle this? How do we secure supply before our competitors in the upstream tiers of our supply network? Keep in mind too that even nonmanufacturing companies are finding value in advanced supply analytics.

Q: What is the delivery model for optimization technology at this time? Is it very consulting-intensive?

A: The model today is highly services-based. It follows the same basic trend we have seen with advanced supply chain analytics, finite scheduling or supply network design applications. You start with some really smart operations research people (read, mathematicians) in a lab. They build some cool algorithms and put them into the hands of the advanced consulting companies. These hybrid software/services companies take the solvers out to their client bases and figure out how to apply them in real world situations. They usually then bring ideas back to the developers who turn these tools into analytic tools that are embedded into larger packaged applications or bought by these larger providers outright.

What you want right now is to be working with a technology provider that offers a great deal of domain expertise in terms of how the analytics software works and how it solves these problems. You want domain experts, the operations research hot shots, to be sitting down with your leading minds in supply chain and procurement. That is where people really talk about what the problems are and work together to solve them. That is how an organization can learn to use the tools as they're meant to be used and make sure they learn how to realize the benefits--if at all. The biggest thing to be careful of is seeking the silver bullet when organizationally, you shouldn't be playing with guns. Advanced technology in the hands of companies who aren't ready for them is a tried-and-true recipe for disaster. I won't even begin to name names here.

Q: What are the major obstacles that providers of optimization technology for sourcing and supply chain must overcome?

A: First, they have to get companies to recognize there are problems, that they could be making better sourcing and procurement decisions. There are so many tactical problems competing for people's attention, there's little time left for thinking about how you're going to get ahead of the game. In this very noisy technology and services market, it's difficult to get people interested in these cool little tools.

A second big problem is that the return on investment (ROI) for analytics is not always explicit. Very often, the ROI is based on cost avoidance in a climate where everyone is focused on achieving quick-hit hard cost savings. The business case for analytics requires senior supply leadership to recognize the value of making supply chains more effective and also to be capable of executing the changes that really harvest the savings opportunities the analytics engines uncover. You can do a really great strategic supply contract, but you still need to have the compliance. That's why analytics technology/service providers will often use gain-sharing arrangements.

The really strategic analytics find opportunities that are at the supply network level--not at the local level. Even great techniques like lean manufacturing are often sub-optimized. What this means is that you have to be ready for cross-functional and cross-enterprise collaboration to optimize the whole network--even if it means hurting the metrics of certain functional silos.

A third problem for the analytics companies is that so many big ERP and supply chain companies already dominate the technology landscape. CIOs don't like to hear about adding new software providers. You spend your time justifying why your ERP provider can't do it. The reality is that specialized analytics are very good adjuncts to the big ERP systems and data warehouses that most everyone is deploying in some form or fashion.

Q. Is there danger in applying rocket science where simpler tools will suffice?

A: You don't have to use optimization for most problems. There are plenty of very good basic analytics (for example, spend analysis) that don't use optimization. But for a few big problems, you want to bring out the big guns. Tactically, the analytics allow you to speak with data. They give you the ability to say, for example, 'Here is what X or Y business roles are costing us. It gives you a really clear way of understanding the costs associated with decisions or assumptions you have made in the past. Maybe you have a role that says you will use two suppliers for a particular commodity. An exercise in bid optimization may show that you would save $8 million by expanding to three suppliers. That would far outweigh the administrative cost of adding a third supplier. Optimization can also help you make cases for more difficult change initiatives like reducing parts, firing unprofitable customers or outsourcing manufacturing and logistics functions.

Q: It has been argued that optimization technology can be dangerous - even lead you astray - if deployed too narrowly, say, within a function rather than holistically in a corporate organization...

A: That really depends. With bid optimization, the modeling variables are generally under the control of purchasing or cross-functional teams that include people from engineering, logistics, quality and manufacturing. That ensures all the necessary requirements get expressed in the model. With supply chain optimization, the value certainly increases when you extend the model across the supply chain, including suppliers, so you can do increasingly advanced tradeoff analyses. A key thing is to get everyone on board with the same measure of success. A great example is Motorola's complexity metric. Most everyone is on board with that metric; and they understand that reducing platform, product and supply complexity is going to reduce inventory, cost and customer dissatisfaction. Once you get everyone on the same page in terms of the metric, you can use advanced analytical tools to figure out how you will move that metric in the direction you want. Really, we're talking about a much bigger toolbox for identifying opportunities - opportunities that take you beyond basic spending analysis techniques to be used for supplier leverage in a commodity, and more towards hitting the other sets of value levers like demand management, specification management and value re-engineering.

Discuss this:

Your email address will not be published. Required fields are marked *