Category Archives: Decision Optimization

Optimization Backed Sourcing Platform … Or Bust Part III


This is the third part of a five part series that revises and ties together key ideas outlined last year on Sourcing Innovation across multiple posts. Regular readers will be familiar with much of the content, but the integrated perspective should help to cement the ideas in regular readers and new readers alike.

This post is largely based on It’s Not Optimization, It’s Strategic Sourcing.

In our last two posts we outlined a complex scenario that could not be accomplished with a traditional sourcing suite that was just a loosely coupled set of modules that did basic sourcing tasks and provided many reasons why the power of the suite did not even come close. Simply put, we have not only reached the point where it is impossible to define a sourcing event of any magnitude without hitting at least a few of the nine dimensions of complexity covered in “what defines complex sourcing and why does it matter” on Spend Matters but we have also reached the point where the data collection, manipulation, and analysis requirements are so intensive that only a sourcing solution built on, and backed by, a true optimization engine is going to be able to handle the data, manipulation, and analysis required.

Now, we’re not saying that the right strategy for every event is optimization, but we are saying, as per SI’s already classic paper on Optimization, What Comes Next, that we have reached the point where you cannot determine the right strategy without optimization to at least build and solve a baseline cost model given current market prices and expected bidder increases or decreases from the last event to determine whether or not optimization might be helpful.

For example, while a 3% savings potential might be enough for a (strategic) sourcing auction or optimization-based multi-round RFX, a 3% drop in expected product cost does not necessarily imply a 3% savings potential. If that drop is from remote suppliers that ship down lanes where costs have risen 10% and shipping is 30% of the overall total cost model, there is likely no savings potential. The right strategy is likely a renegotiation with the incumbent for a contract extension or a spot market buy. Similarly a 2% drop in price combined with a 5% drop in logistics costs could equate to a 3.5% savings potential under the right circumstances, which is substantial on a 50M+ category.

Plus, with bundled discounts, volume discounts from suppliers and carriers that take effect at different price points, different import and utilization costs for each supplier, and an ever increasing plethora of capacity constraints, mandatory award splits to minimize risk, secondary goals of minimal environmental impact, and so on, it’s often impossible to determine what the lowest cost solution is and, thus, if the cost increase associated with assigning a (greater percentage of the) award to a preferred supplier seen as being more valuable in the long term is actually worth it.

In many situations, there’s just no way to do a strategic analysis and justify a strategic decision without a basic level of true mathematical optimization capability that can take all costs and constraints into account. Spreadsheets were breaking under the strain of basic sourcing requirements years ago. Now these sheets are just shards of glass — which will eventually cut you if repeatedly handled.

That’s why you have to not only graduate from a suite to an integrated sourcing platform but, when you do so, select one with integrated optimization capability. While you won’t need to use optimization in every event, you’ll always have the option and always be able to use the advanced mathematical capability to determine both the savings potential and, sometimes, even the odds of success (as you will be able to iterate through dozens of what-if scenarios based upon expected supplier and carrier bids and proposals).

But while we have clarified why traditional suites, built from a set of loosely integrated modules, are not modern sourcing platforms ready for complex sourcing, we have not clarified why many of these suites cannot be upgraded to sourcing platforms. We will address this in our next post.

Optimization Backed Sourcing Platform … Or Bust Part II


This is the second part of a five part series that revises and ties together key ideas outlined last year on Sourcing Innovation across multiple posts. Regular readers will be familiar with much of the content, but the integrated perspective should help to cement the ideas in regular readers and new readers alike.

This post is largely based on It’s Not a Suite, It’s Just Sourcing, Part II.

In our last post we made the rather bold claim, which is probably going to irk a lot of vendors, that it’s NOT a Suite, It’s JUST Sourcing. SI likes vendors that are trying to build solutions to solve their customers’ pain points, and has chronicled the efforts of many over the years, and thus isn’t doing this series to be irksome. SI is doing this series because it’s not 2005 anymore, it’s 2015 and the nature of, and need for, Sourcing has changed as global trade has become more complicated, supply chains have lengthened, risks have increased, and sourcing has become more complex. Today, sourcing absolutely has to be more strategic and Suite Sourcing is NOT Strategic Sourcing. In today’s post, we’re going to begin to clarify why.

Our last post outlined a hypothetical, but realistic, example in the high-tech space, discussing a typical, primary, sourcing event for a company that assembled custom-built high-end workstations for software developers and engineers. We started by discussing the primary factors that the Sourcing analyst was likely to identify as well as two strategies the analyst was likely to take. This led to a perceived event progression and a plan that looked like it was easily executable in you average modular sourcing suite. We did this to make it clear why many companies fall for the fallacy that you can attack sourcing in a step-wise fashion using a modular suite, and, as a result, why some vendors still believe that a modular suite is the way to go. The reality is that, at a quick glance, it does look like this is the right approach and that there is no reason to question it — even though there is a big reason. Namely, the approach is wrong.

The reason being is that, in reality, the event is not going to go as planned.

Specifically, it will not be an analysis followed by an RFP followed by a single auction / optimization analysis followed by a push into the contract management system. One or more, with emphasis on the more, of the following will happen:

  • the RFX will come back and some of the requested bid fields will be empty because the supplier is no longer producing the product
  • the RFX will come back and there will be new products that the buyer did not know about with new bids (and new interdependencies to be mapped)
  • the logistics carriers will come back with quotes much higher than expected and/or a logistics carrier or 3PL will withdraw (due to overcommitments) and lanes will vanish
  • stakeholders or key customers will change requirements post RFX issue and you will have to go back and ask for prices on next generation products, which might still be in final design stages
  • the baseline optimization will come back with completely unexpected results and once the analyst uses spend analysis to dive in, the analyst will find a number of outliers in the incumbent bid and realize that she has to go back and ask for verified or corrected data
  • the auction will end with three suppliers almost equal on baseline scoring and extensive analysis will be needed to determine which supplier gets 50%, which supplier gets 30%, and which supplier gets 20% in the 50/30/20 split dictated by the stakeholders to minimize risk

In these situations, respectively

  • the analyst will have to identify a larger supply base and send the RFX to more suppliers
  • the analyst will have to research the new products and decide whether to accept them or not and then, possibly, ask the supply base to bid on (comparable) products in a revised RFX
  • the analyst will have to invite more carriers to bid and consider alternate lanes, possibly from secondary (air)ports to secondary (air)ports
  • the analyst will have to create revised specs and go back to the supply base for additional prices and options
  • the analyst will have to backtrack to the spend analysis step on the submitted data, followed by a request for bid verification and a repeat of the optimization on revised data
  • the analyst will have to go back to the analysis step to identify which bid components were strongest for each supplier and then compare that to existing supplier scorecards (to determine likelihood of on-time delivery, quality guarantees, price consistency, etc.)

In other words, the event is not going to go as planned and it’s not going to be a sequential progression from analysis to RFX to auction/optimization to award. Moreover, most events are going to see multiple occurrences of the above hiccups and require an almost random workflow that uses all of the sourcing capabilities of a suite multiple times.

Moreover, the transitions back and forth will need to be seamless. If an analyst has to push data out of the optimization “module” into the “analysis” module for detailed data and outlier analysis, then push the data, with insights, back into the “RFX” module for revised RFX data collection, and than push the revised RFP data back into the “Optimization” module for revised analysis only to find out that the lane cost is coming out higher than expected in the preferred award, indicating that there is still an additional opportunity if logistics costs can be lowered, then this “modular” workflow quickly becomes a nightmare.

Plus, in this situation, the analyst will have to do an in-depth analysis of the logistics cost to determine if costs can be lowered simply by inviting more carriers to bid, analyzing primary and secondary lanes, or doing something progressive like using the organization’s sourcing expertise to help a provider lower their overhead with better insurance rates, communication plans, and office & computer supplies from the organization’s GPO contract. Then, after this analysis has been done, which will likely take the form of multiple what-if optimizations using various cost models, the analyst will have to go back to the RFP, issue the revised RFP with more options to current and new suppliers, push the data back into the optimization module and continue.

In a modern sourcing project, one cannot separate data collection from cost modelling from analysis from bidding from optimization — it is all one integrated sourcing process that lathers, rinses, and repeats until the solution is found and the event is done. And any provider that thinks you can separate pieces out and take a modular, piecemeal approach and build up to a suite, one module at a time, is still living in 2005 and should be approached with caution. It’s not a suite, it’s just sourcing. And, as indicated in our previous post, and as will be discussed in more detail in a future post, it’s not optimization, it’s strategic sourcing.

Optimization Backed Sourcing Platform … Or Bust Part I


This is the first part of a five part series that revises and ties together key ideas outlined last year on Sourcing Innovation that were spread across multiple posts. Regular readers will be familiar with much of the content, but the integrated perspective should help to cement the ideas in regular readers and new readers alike.

This post is largely based on It’s Not a Suite, It’s Just Sourcing, Part I.

For a while now, Sourcing Innovation has been effectively saying that if you do not have an optimization-backed sourcing platform, you’re not ready for the modern era of complex sourcing. And SI means it. This isn’t to say that you can’t get value from a modern suite that covers the end-to-end sourcing lifecycle, or that you can’t get value from a first generation optimization platform, because you can — especially if you haven’t had these solutions before. However, every last-generation solution has a limit on the value it can deliver. Some of these limits are low, and some of these limits are quite high — so high, in fact, that it can take years, and sometimes a decade or more, for the average organization to hit the ceiling. But once that ceiling is hit, the organization has to know what comes next to continue extracting value from the supply chain. So this is a post about what comes next for the average organization and, most importantly, what comes now for the leaders who have already realized that their first generation optimization modules and / or first generation suites are failing to deliver the value they need today.

The reality is that, these days, Sourcing needs to be much more strategic and is thus not an activity that can be accomplished as a discrete set of loosely connected tasks where you can pick and choose what you need ahead of time. Strategic Sourcing is an activity that needs to both analyze the need and the market situation and respond to the stimuli the market is providing in a dynamic fashion.

This can not be done according to a pre-planned, limited set of tasks. To clarify, let’s take a hypothetical, but realistic situation. Let’s say that the company is a high-tech retailer selling custom assembled high-end development boxes to software development and engineering shops. This company will not be buying pre-configured Dell and HP machines, targeted to the consumer market, but custom configured boxes using high end motherboards, which may be manufactured by the same production houses that manufacture boards for companies like Dell and HP, high end Intel and AMD processors, ultra-fast high density DRAM, high-end solid state drives, mid-tower cases with extra fans, etc.

This might sound like a relatively easy sourcing event as there are a relatively small number of acceptable motherboard manufacturers, DRAM manufacturers, drive manufacturers, case manufacturers, and only two chip manufacturers, but even 5 * 5 * 5 *5 * 2 = 1350 and each manufacturer might have over a dozen acceptable options — and it’s hard to say up front how many of these combinations are not only viable, but acceptable (as even though it might be feasible to connect the components, there might be driver or other issues that affect compatibility or performance). In addition, new manufacturers arise once in a while and old manufacturers fail or sell out. Last year’s customer spend pattern is not the same as the spend pattern two years ago, and until year-over-year is analyzed for multiple years, you have no idea of the average deviation.

In other words:

  • you may or may not need a pre-event spend analysis to determine potential volume leverage points, the opportunities with supply base consolidation, and expected savings potential, all depending on when the last event was run, how much data you have, and current market data points
  • you may or may not need optimization; if you restrict the bid to pre-configured systems, because business is up 40% and you need a quick event to get through the rest of the year with plans to do a more detailed analysis in 6 months, you can probably get away with a weighted auction, but if bid options are open, you will probably need optimization to handle all the data
  • you may or may not need multiple RFX rounds, so you may or may not need a supplier portal to handle the communication necessary for a multi-round event

And this is all fairly obvious, so you are probably thinking

  • if I need the analysis, I invoke the spend analysis module, get my insights, and plan my strategy
  • then I invoke the RFX module to create the RFX
  • if I am doing multiple rounds — I have to configure the Supplier Portal instead of just sending out the Excel spreadsheets (which I would import on return otherwise)
  • when the data is retrieved, validated and cleaned up then I either
    • push it into the auction for a weighted auction or
    • push it into the optimization module for optimization-backed analysis
  • when I have my winners, I push the data into the contract management module for draft contract creation

… easy-peasy, right?

Wrong!

This is the real world, and it never works this way, as we will discuss in our next post.

RiverLogic: Bringing Optimization to the Enterprise

In our last two posts on Beyond Sourcing Optimization we noted that Strategic Sourcing Decision Optimization (SSDO) is just one area where optimization can be successfully applied in a progressive organization that is a leader in its industry. An average enterprise organization is ripe with opportunities for the application of optimization technology — optimization technology which can easily add up to 5% to the bottom line if properly applied.

RiverLogic, which is advertising prescriptive analytics technology and enterprise optimization, is one such company that is tackling enterprise optimization. Advertising holistic decision support that performs simultaneous optimization of the entire business model, the optimizer is capable of integrating demand, production, and inventory optimization into a single holistic optimization model that, when run, optimizes production and inventory against demand to maximize profit by optimizing revenue against production and inventory (carrying) costs.

As one may have gathered by our previous post, this is no easy feat. Production optimization requires one type of model that understands production line throughput, machine utilization overhead costs per hour or unit, associated workforce requirements, associated costs during regular and over time, raw material inventory costs, and logistics costs at different production levels. Inventory optimization is a different type of model that must take into account the myriad of costs that contribute to the amortized inventory carrying cost and that include, but are not limited to, warehouse overhead costs, labour costs, and depreciation costs and balance these against logistics costs from more or less frequent orders. Finally, demand optimization is its own beast as one has to have a relatively good understanding of the different types of marketing spend and how each will influence market demand, the costs of production at various volume levels and delivery commitments, the associated lifecycle costs including outbound distribution costs, warranty and service costs, and any end-of-life reclamation costs (if one or more locales in which the product is being sold have mandatory reclamation or recycle laws for the product or one or more of its components). Now, one can create a mega-model that encapsulates inventory and production costs into the demand optimization, but it’s not easy, and that’s why few companies have tackled this problem (just like few companies have tackled true SSDO). And most that have tackled this problem do so by building custom models for their clients that require individuals with advanced degrees in Mathematics or Operation Research to run.

However, the RiverLogic platform, like leading SSDO platforms, comes with this model “out-of-the-box” and all a user has to do to build a basic model is get the data. At this point you’re probably thinking this is a show-stopper as

  • the amount of data required to populate such a model is extremely extensive and
  • outside of the ERP, no one system has even a fraction of the data required.

RiverLogic understands this perceived dilemma as well and that’s why their platform integrates with over a dozen major ERP and Accounts Payable systems because when you get down to it, that’s where the majority of the cost data required for a holistic demand optimization model, that simultaneously balances inventory and production, resides. Once the proper integrations are done, the model can be run out of the box and the organization can instantly see relative to its demand forecast the optimal production and inventory levels (and, as a bonus, the optimal distribution plan and cost model as logistics are also accounted for in this holistic model). This will allow an average organization that has not simultaneously balanced these models before to shave at least an extra 5% to 15% off of overhead costs and, if one or more of these models haven’t been run before, even more. And these savings will trickle down straight to the bottom line the instant the plans are updated.

But the power of the platform doesn’t stop there, like the best SSDO solutions, it also supports powerful what-if optimization that allows the organization to see how production and inventory plans change if the demand projections were to change, if more money was allocated to marketing, or if the estimated impact of marketing campaigns were more or less successful than initial predictions. This model can be run any time and plans updated dynamically, taking effect with the next (automated) order upon publication to the demand / inventory / order management module of the ERP(s) that the platform integrates with.

Now that leaders like you are using decision optimization in your Sourcing and advanced spend analysis in your planning, you’re ready to apply that knowledge and capability across enterprise operations and, as such, are ready for the enterprise optimization (and prescriptive analytics) that innovators like RiverLogic are offering. RiverLogic is one of the handful of companies you’re going to be hearing a lot more about in the coming year
and one that should definitely be on your radar as you look to take cost control across the enterprise (because what good is saving 10% on a category if poor operations just eliminate that savings after the deal is signed?).

Beyond Sourcing Optimization: The Best Bundle is Only the Beginning Part II

As per our last post we recently discussed the criticality of optimization in
It’s Not Optimization, It’s Strategic Sourcing, explained that Even “Simple” Categories Hide Extreme Complexity, and pointed out Why Your First Generation Platform is Not Ready for Modern Sourcing in the hopes that you would understand that you need to be ready for Complex Sourcing.

But Strategic Sourcing Decision Optimization (SSDO) is only one area where optimization can be applied to add organizational value. There are at least half a dozen other areas where optimization can be successfully applied in a progressive organization that is a leader in its industry. As noted in our last post where we briefly discussed inventory optimization, production optimization, and demand optimization — three areas that can hide considerable cost savings when properly analyzed — there are a number of areas in an organization where optimization can identify considerable value. Today we will discuss three more areas.

Service (Level) Optimization

The goal of service level optimization is to find the right balance between customer satisfaction and service cost to maximize profitability while minimizing customer dissatisfaction. While everyone would like a robust high-quality product that lasts until they are done with it, that’s not always feasible. The very nature of many product lines — electronics, automotive, plastic-based CPG is that the products will break down with regular use and/or regular exposure to the elements. Better materials, better manufacturing, and better care will lengthen lifespans, but computing equipment, cars, and plastic boxes don’t last forever. And if the intended lifespan is 3 years, it’s not cost effective to build a product expected to last 13 years. However, the nature of things is that if you build a product expected to last 3 years, some units will breakdown and need to be replaced before the 3 year mark is up (and some will last longer). So what level of warranty do you need to offer, what level of service (in terms of repair / replacement window), how much will it cost, and how much will the market bear. Finding the right base / extended offerings and price points is key to maximizing both consumer demand and customer satisfaction.

Asset Optimization

Big organizations have a lot of assets, often assets they often don’t know that they have. For example, unused compute power in the data centre, unused production time at the factory, and unused equipment in the yard. This last category can be a huge burden to an organization that is paying a lease or amortized monthly payment on a 10-year plan for equipment that is only being used part of the time. (That’s why renting by the job or renting out might be a better solution.) And if an organization is continually renting the same equipment for multiple projects, it might make sense to buy and share the equipment between projects, even if it has to be transported between locations. Asset optimization can save an organization a lot of money and can often, when done properly, considerably increase working capital. This brings us to:

Working Capital Optimization

Optimizing the balance between assets and liabilities, working capital ensures that a company has sufficient cash flow in order to meet its short-term debt obligations and operating expenses without creating a crushing debt load or jeopardizing long term profitability. Working capital optimization is tricky as it involves balancing with DPO (days payable outstanding) with DSO (days sales outstanding), early payment discounts (on the inbound and outbound supply chain), supply chain and invoice financing, short and long term investment opportunities, and short-term gains vs. long term cost reductions. It’s not easy, and, like all of the other examples of optimization covered in this brief two-part series, often requires sophisticated optimization.

These are just a few of the examples where optimization can yield significant benefits beyond sourcing and where Sourcing can bring additional savings and value to the enterprise since it will be able to collect a lot of the data and intelligence that is required to build, and solve, the sophisticated models required.

In future posts we will discuss these types of optimization in depth as well as the new breed of providers tackling these types of supply chain optimization. Stay tuned.