Category Archives: Decision Optimization

When It Comes to Optimization, You Need Every Insight You Can Get!

Even though it’s been almost a decade since Strategic Sourcing Decision Optimization (SSDO) has been not only readily available, but affordable (especially when one considers that back to back Aberdeen Studies in the noughts demonstrated that advanced sourcing, which is based on optimization, saved an average of 12% per event, which means that companies that employed optimization on large categories often saw an ROI after their first event), most mid-size and larger companies aren’t using it. In fact, most mid-size and larger companies haven’t even tried it!

Why is this? There is a laundry list of reasons, but the most important are probably:

  • misinterpretation and misinformationThere is still a lack of understanding about what optimization is and how important it is to your strategy sourcing efforts. A lot of people believe that optimization is only for the largest categories, the most complex categories, companies with complicated manufacturing supply chains, etc. This is not true. Optimization is relevant to every strategic sourcing project, small and large. The only question is how important is it — does the event revolve around the optimization or does the optimization revolve around the event?
  • fearBecause it’s misunderstood and, more importantly, because it is math, it is feared. (It’s important to remember that less than 1 in 7 American adults are “proficient” at math. This means that while your senior analysts with a strong Operations Research (OR) background will be hesitant of optimization, your average buyer will be, to borrow a colloquialism, scared sh!tl3ss. And, unwilling to admit this fear, he will do everything he can to come up with dozens of excuses as to why optimization is not applicable to your problem or why other methods will perform better.) And moreover, because of the misinformation out there which doesn’t tell you that the good solutions handle all the math for you, and all you need to do is specify the demands and the constraints (and their priorities if not all constraints can be simultaneously solved), people avoid (strategic sourcing decision) optimization when they should be embracing it.
  • costOptimization solutions used to be expensive. Very expensive. Back when there were only a couple of known solution providers (in the e-CHAOS pack), and sourcing suites started in the six figures, optimization solutions, even for a single event, were six figures, and sometimes seven for unlimited use. If you weren’t guaranteed of a high six figure return off of your first event, and a high seven figure return over the course of the year, this was a big risk to take. But that was then, and this is now. Today, optimization solutions start in the lower end of the five figure range, and unlimited annual licenses start in the lower end of the six figure range. And their power and performance is at least ten times what it was a decade ago. Models that used to run for hours now solve in minutes and an analyst can run dozens of what-if scenarios in a day, quickly getting to the best price-value trade-off for the organization.

So how do we get optimization into the hands of the masses, and more importantly into your hands (if your colleagues are holding your organization back)?

We deal with the roadblocks we discussed.

How do we deal with the roadblocks?

We start with education. We educate people that they don’t have to be a math whiz (because the math whiz is only needed to build the solution, not to use it), that a strategic sourcing decision optimization solution isn’t hard to use, that it doesn’t cost a lot, and that it does generate a return. And we hit them on all fronts. Third Party, Provider, and Practitioner.

To date, it’s been mainly third party, and, unfortunately, mainly SI spreading the message of optimization. But now we have a few providers working hard to spread the message as well. BravoSolution, who has been kind enough in the past to sponsor SI to help with this effort (and who offered you an Illumination on The Future Of Optimization) has been working hard to spread the messages of Optimization, Analysis, and the integration thereof in what they call High Definition Sourcing for a few years now. A new provider in the SSDO arena, and the first new provider to provide a true SSDO solution since Iasta back in the 2007-2008 timeframe, that we’ll announce shortly, is also taking up the challenge.

And Trade Extensions, who has also been kind enough to sponsor SI to help with this effort, and who has also been providing industry leading optimization solutions and education for a few years now, has just doubled down on the education effort, starting with a new INSIGHTS series focussed entirely on optimization. Consisting of a series of nine interviews with Founder, Chairman, and Optimization Guru Arne Andersson and CEO, Freight Trader, and Master Buyer Garry Mansell, this series will attempt to burn away the fog on optimization, make it a standard part of your sourcing suite, and lay the foundation for a series of follow-up educational offerings which will include white-papers and webinars on the subject.

Because optimization is for everyone, not just the 1%!

Are You Doing It Wrong?

If you’ve been following the media, you know that we have reached a point were most major business publications are now putting focus on Supply Chain as your top risk and your top opportunity.

You also know that these same publications, and the solution providers that follow, and reference them, have been preaching the following solutions to not only tame the risk but increase the opportunity.

Comprehensive Category Management

Spot buying individual categories at market lows or evening running reverse auctions at opportune times is not category management. And for that matter, neither is an event that covers the entire category. At this point you probably think that the doctor is losing it a little, because how could it not be category management if you are addressing the whole category?

It’s Simple. Category Management isn’t just about grouping all seemingly related items and running an event, it’s grouping items that have related characteristics that allow the items to be sourced effectively under the same strategy. For example, while it might make theoretical sense to group printers, ink, and paper together — because you use them together, from a sourcing point of view, ink and paper often go better with office supplies and printers with hardware. You can probably get them thrown in for free with a server purchase. But that’s just the start. If you source a lot of metal parts, you should probably group them by primary metal, since the price of steel, aluminum, etc. will largely dictate their prices and it might even make sense to not only source all of the parts from the same supplier but even buy the metal on behalf of the supplier with your better negotiating power and/or credit rating.

Supply Chain Risk Monitoring

Natural and Man-Made disasters devastate supply chains when they result in raw material or product unavailability for weeks or months. When a company doesn’t understand their dependence on a single source or the risks that single source is subject too, they can figuratively get caught with their pants down to say the least.

As a result, most leading companies in the Risk Management arena are now tracking and monitoring their tier 1 supply base for not only missed deliveries, but late shipment dates and inquiring immediately when something is late shipping. However, by the time a shipment is late, it’s often too late to go to another source if the reason for the lateness is the lack of an important raw material. So the smarter companies also ask their suppliers to let them know when their suppliers miss a delivery. This is better, but sometimes this is still too late. You need to track the primary sources of the raw material and their ability to produce. Not only the companies, but their locations. All natural and man-made disasters in the region and then evaluated for impact and if the producer of the primary raw material or part is potentially at risk, they make sure, or ask their tier 1 supplier to make sure, that the raw material or product can still be delivered on time and if it can’t, these leading companies immediately seek a secondary source (or lock up available supply pre-emptively) — not two weeks after the tier 1 supplier required the raw material to meet the commit date.

Big Data

The only buzzword on par with big data is cloud. According to the converted, or should I say the diverted, better decision are made with better data — the more data the merrier. This sounds good in theory, but most algorithms predict demand, acquisition cost, projected sales prices, etc. based on trends. But these days the average market life of a CPG product, especially in electronics or fashion, is six months or less, and the reality is that there just isn’t enough data to predict meaningful trends on. Similarly, every disruption impacts the cost, and these disruptions are as unpredictable as future sales predicted using trend models with insufficient data.

You use all of the data available to validate your operations, procurement, and financial situation. Not to blindly predict future sales or prices. An over-reliance on big data is often more dangerous than not having data at all.

Anticipatory Demand Planning is Good, but Anticipatory Shipping?

SI can believe that Amazon patented a Method and System for Anticipatory Package Shipping (US Patent 8615473) but can’t believe it would use this for more than a small number of items. Nor does it believe the system would be implemented as outlined in the patent as filed, at least in the short term.

It took Amazon 7 years to turn its first profit, and while Prime is currently very profitable to Amazon (which makes $78 more in profit per year per Prime customer, on average, than non-prime customer according to CIRP’s market research – Source: Wired), those margins would drop substantially if Amazon started shipping tens, or hundreds, of thousands of packages a year that no one wanted. Amazon does have an efficient distribution network and probably has the absolute best deals with postal and courier services that can be papered, but every shipment costs money and every unnecessary shipment eats into profit. Returns cut into profit margins enough, how much are returned shipments to nowhere going to cost?

Thanks to big data, predictive analytics is getting better by the day, but it’s still hit and miss at a granular level. While it’s pretty easy to use correlation data across a large customer base to predict that you are likely to desire an item, it’s harder to predict whether or not you’d actually buy it, and if you would, at what price point, assuming you don’t already own the product in question. (It’s always telling the doctor he wants books and media he already owns.)

As a result, any predictive analytics at the individual consumer level are going to be hit-and-miss at best. Predictive analytics work best across a large consumer base with a lot of data where one can predict that, on average, 5 in 100 people who match a profile will buy the product from Amazon.

And, from Amazon’s viewpoint, the best use of the predictive analytics is on new releases, as the bulk of sales in many of its categories, and books and media in particular, are in the weeks immediately following a new product release. With the right data and the right algorithms, it can not only predict how many units it is likely to sell against its current customer base, but if the demand is enough, how many in each region that is associated with each distribution center and how the orders will likely track over time on a daily basis.

In this, and only this situation, would anticipatory shipping, and in particular, anticipatory packaging, make sense in the short term. For example, if Scott Adams were to release a new Dilbert book and Amazon predicted 200,000 copies would be sold in the first 3 weeks, and expected that it would get 50,000 of those sales, pre-packaging 40,000 for shipment and then distributing those across it’s DCs such that each DC received a number of books proportionate to the expected sales in the serviced area would be a good idea. All Amazon would have to do to speed up shipment would be to slap the delivery address on the boxes as the orders came in and have them ready to go in the next pickup for local delivery.

In the future, once the system is fine-tuned and its delivery partners have the technology to replace a unique delivery address identifier with a specific delivery address on-the-fly, Amazon can pre-ship a set number of these pre-packaged items to the local post office or delivery company every day, which can, in turn, load those packages onto the appropriate courier truck each morning as the addresses in the system are updated with consumer delivery addresses sent over by Amazon upon each purchase.

But not everyone would get faster shipping service. In order to prevent too many unnecessary shipments and loss, Amazon would have to err on the side of caution and pre-package (and pre-ship) less unit of an item than it expected to sell, and restrict the anticipatory shipping and packaging to only those items expected to have a large sales volume. In most cases, the best Amazon will do is optimize the distribution of inventory across its warehouses. However, this can still take a day (or two) off of average delivery time, so this is still a good start.

Any differing opinions?

Where Is Your Greatest Risk? Not Where You Think It Is.

As per a recent piece by Simchi-Levi, Schmidt, and Wei in the current issue of the Harvard Business Review on managing unpredictable supply chain disruptions, there is little correlation between how much a firm spends annually on procurement at a particular site and the impact that the site’s disruption would have on company performance. In reality, the greatest exposures often lie in unlikely places.

Moreover, in many supply chains, these exposures are typically not realized until a low-probability, high-impact event — such as a Hurricane, Earthquake, SARS outbreak, or other mega-disaster — occurs. In these situations, companies find out that they significantly underestimated the impact and are not adequately prepared because their traditional models for evaluating and preparing supply chain risk break down as there is typically a lack of historical data for low probability, infrequently occurring, high-impact events. (Big companies have to deal with poor supplier performance, forecast errors, and transportation breakdowns everyday and traditional risk models can thus adequately predict, and allow the organization to prepare for, these impacts.)

But, as the authors point out, it doesn’t have to be this way. Companies can not only determine the potential magnitude of a disruption without historical data, but can even do so without even knowing what the disruption is. This is because, at the end of the day, the specifics of a disruption don’t really matter — only its impacts do. Be it flood, famine, or fire — you don’t care why your factory isn’t producing — you only care that it isn’t and you have to find an alternate source of supply. And it is possible to model the impact of a disruption at any point of your supply chain without knowing the event that caused it, as an impact is either going to eliminate or cut off supply or production.

To this end if, as the authors indicate, you develop a mathematical model (that can be computerized) that focuses on the impact of potential failures at points along the supply chain (such as the shuttering of a supplier’s factory or the inaccessibility of a distribution center), rather than the cause of the disruption, you can quantify what the financial and operational impact would be if a critical supplier’s facility were out of commission for, say, two weeks — whatever the reason. And that’s what you really care about.

In their paper, the authors describe a sophisticated linear optimization model that integrates predicted Time-To-Recovery (TTR) factors for each node (based upon historical recovery times for the supplier or distributor after a disruption) with Bill-of-Material (BoM), operational measures, financial measures, in-transit inventory levels, on-site inventory levels and demand forecasts for each product. When one node is removed at a time from this model, it can be used to find the supply chain response that would minimize the performance impact of the disruption (such as reducing inventory, shifting production, expediting transportation, or reallocating resources) and then calculate the resulting operational performance impact (PI). The node with the largest PI presents the greatest risk and is assigned the largest risk exposure index (REI) of 1.0 (and all other nodes are indexed relative to this value).

While you may need such a model to determine the full impact of a disruption, you don’t need such a complex model to determine the big hidden risks in your supply chain (which are often the result of sole-source supply arrangements somewhere in the supply chain, possibly at tier two or three). All you really need to do is map the full supply chain for every product you produce down to the raw material supply. Then you can quickly identify sole-source supply, single-factory or single location production, bottle-necks in the distribution network, etc. which lead to hidden risks.

And once you have identified the major risks, and collected the data to appropriately access the potential impacts of a disruption, you can build local models to analyze the extent of the risk exposure. And as you build more and more models, you work your way up to the point where you can begin working on the model described by Simchi-Levi, Schmidt, and Wei, incrementally. No big bang modelling approach needed. All you need to do is get underway with a good supply chain visibility solution, such as Resilinc‘s.

Bravo Business Center 2.0 – A Complete Category Solution for Retail Part III

As per part I, two years ago we reviewed BravoSolution’s Business Center Category Sourcing Solution that took e-Sourcing to a new level for nine common categories that provided the Supply Management organization with a considerable sourcing challenge. In addition, we noted that BravoSolution didn’t stop there and kept going until they built a solution that, capturing the years of experience and knowledge built up by their global sourcing and solutions teams (who work out of offices in ten different countries on four different continents), captured all of the common categories for entire industries. This allows a sourcing professional in those industries to use the Business Center as a complete sourcing solution and apply built-in best practices built up from decades of experience.

Then, in Part II (dot 1 and dot 2), we noted that one of the industries that the Business Center serves, out of the box, is MRO because it is a vertical that is almost tailor-made for a business centre solution. Even though, as a category, it is one of the broadest categories imaginable, MRO organizations are generally not sourcing any particular product or service in volume and success often depends not on identifying the supplier who can give you the best price at the best service level on a part, but on identifying the supplier who can give you the best average price at the best average service level of a large market basket of parts (or the supplier who can bundle the services associated with installing a related market basket of parts at a competitive rate). Part II detailed how the business centre guides a buyer through the process, automates as much as possible, and makes it as easy as possible for a buyer to take a sourcing event from conception through award.

Today, we are going to discuss the business centre solution for Retail, used by some of the largest retailers in North America and Europe. The retail solution is designed to support categories with a large number of products that need to be sourced to a large number of distribution centres which serve a large number of stores that need variable volume levels of different products. These events need to be built on sophisticated models that can fed to an optimizer because the variable demand for a product means that not only do you need to consider multiple bids from multiple suppliers and multiple lanes, but buying certain products from certain suppliers for low demand locales could result in a lot of LTL shipments that will significantly increase transportation costs and buying products that will need to be shipped great distances for repairs or warranty claims will also drastically increase TCO.

The BravoSolution Business Center Solution for Retail allows the sourcing organization to define all of it’s distribution centres, all of its stores, the stores served by each DC, markets, the markets served by each store, the warehouses for each supplier, and the DCs that the supplier warehouses are able to serve. It also allows the sourcing organization to define all of the items that it buys, all of the supplier products, the mapping from supplier products to buyer items, categories for its items, and categories for the supplier items. All of the distribution centres, stores, warehouses, items, products, and categories can be uploaded from an (Excel) datafile, and so can starting prices.

It also allows for the easy definition of a very sophisticated discount model that can capture any convoluted discount the supplier can come up with. In addition to the standard volume rebate by product, volume rebate by spend, volume rebate by category, volume rebate by category spend, and volume rebate by supplier spend, suppliers will often offer new store discounts, co-op discounts, payment discounts, EDI discounts, in-store promotion discounts, defective discounts, and cross-product discounts where a discount will be offered on X for every unit of Y purchased. The workflow, and interface, is set up to allow for easy capture of any, and all, of these discounts.

The workflow also allows for the definition of (additional) item attributes which can be used in qualitative constraints in the optimization model, which will allow a sourcing professional to create models which will only include eco-certified items, validated suppliers, etc. in the award. It also supports price targets, so that a buyer can determine the impact of a proposed price decrease in an optimization model and use this information in fact-based negotiations.

Once all of the stores, distribution centers, warehouses, items, products, and categories are in the system, project definition is extremely easy and, as with MRO, the sourcing specialist is walked through the project which starts by identifying the categories being sourced, verifying the dc-store structure, uploading the projected demand, selecting the suppliers, verifying the supplier-warehouse buyer-distribution center and supplier-product buyer-item mappings, defining any bidding requirements the supplier has to meet, sending out the RFX, verifying the responses, and pushing the responses into multiple pre-defined optimization models, which will include base-line and incumbent models. The sourcing specialist can then create additional what-if models, including what-if models on target pricing, go back to the suppliers for a subsequent bid round, and continue the optimization (and bid-rounds) until the specialist is ready to make an award (and push the award into the contract management module).

As with MRO, the Business Center for Retail is optimized to make sourcing, and re-sourcing, of all of the retailer’s categories as easy and painless as possible so that, if needed, less critical categories can be driven by a junior buyer (under the guidane of a senior buyer) and free up the senior buyer to focus on the high-value and strategic categories. In addition, BravoSolution’s Global Team has the experience to get this solution up and running for even the largest of retailers in a matter of weeks. It’s a quick way for a large retailer to start advanced sourcing and get it’s costs under control.