Category Archives: Decision Optimization

The Basics of Inventory Optimization

Inventory optimization can be defined as the act of balancing supply and demand uncertainty to meet desired services level at a minimum level of investment. In addition to all of the basic factors of inventory management covered in our last post (namely, production, stock, location, transportation, and information), inventory management also considers all of the associated costs — carrying costs, stock out costs, alternate distribution costs, and lead time costs, and tries to balance them.

As such, an inventory optimization solution will allow you to define:

  • your current production & distribution networks, and any flexibility you have
  • the modes of transportation available to you and associated fixed and variable costs
  • the lane options available to you and impacts on transportation costs
  • current and projected demands by SKU, family, and location by time period (month, week, or day)
  • storage and carrying costs
  • desired service levels by SKU, family, and location
  • projected cycle times
  • production capacity constraints and feasible schedules
  • network and storage constraints
  • current contractual commitments
  • upcoming promotions
  • compliance requirements

As well as your flexibility in terms of service levels and working capital and produces optimal inventory recommendations at different trade-off levels. You can then analyze costs and projected losses at 90%, 93%, and 95% service levels (defined in terms of product availability) and make the best decision that balances working capital tied up in inventory and revenue potential.

Share This on Linked In

Algorhythm: Still Pounding Out the Optimization Rhythm on the Tabla (Part II)

In Part I, we re-introduced you to Algorhythm, purveyors of a supply chain optimization rhythm solution platform out of Pune. In the day before yesterday’s post, we discussed their new Inventory Planning Module, inventrhythm, and indicated how it allows you to take your entire distribution network design into account, which is necessary if you truly want to minimize your inventory costs. Then we told you that if you were truly serious about getting the most bang for you inventory dollar, you had to go beyond inventory and also consider your underlying distribution network design, as it ultimately dictates how much your inventory is going to cost you. Just like a bad product design will lock in expensive commodity and engineering costs before it is sourced, a bad network design will mandate higher safety stocks and sub-optimal transportation methods, which will in turn lead to higher carrying and transportation costs. Thus, to truly optimize your inventory, you also have to simultaneously optimize your distribution network to the extent that you are able to do so.

With Algorhythm’s new Strategic Distribution Network Optimizer, which seamlessly integrates their netrhythm supply chain network design module with their new inventrhythm multi-echelon inventory optimization solution, you can simultaneously optimize your facility location, transportation methods, and inventory levels to achieve your end-customer service levels while minimizing your overall inventory-related supply chain costs.

Algorhythm’s netrhythm solution allows you to define the warehouses that are available to you at each level of your network (and to define the warehouses that must be used, or must not be used, in the solution) in addition to source factors and end customer locations; the transportation methods available; the transportation providers available (as well as any that must be used, or must be used, and minimum or maximum business levels); fixed, minimum and/or maximum lot sizes; available lanes, forecasted demand; target inventory levels; and network constraints (with respect to linkages, warehouses, product mix, mode, etc.) and produces a lowest cost distribution network design subject to your constraints that will achieve your target service levels at each location. In other words, it’s a very powerful network design model that lets you take all of the relevant components in your physical network.

But the integrated solution is even more powerful. In addition to the many layers of your distribution network, transportation modes, and logistics providers, you can specify detailed service targets by location, SKU, and period. You don’t have to use average demand levels — you can take into account your detailed forecasts by month, week, and even day. You can model all of your inventory related costs at different demand levels; segment inventory by SKU subgroup, group, and category; and analyze by cluster and channel. You can look at your various cycle times, load factors, and flow options and do so with respect to all of your network and inventory constraints (such as capacity and existing agreements) and cost components (fixed and variable). For example, you can take into account fixed truckload and variable less than truckload rates from a third party and compare that with fixed and variable costs of operating your own fleet (lease, maintenance, etc.). And when you’re done, you get the network design that minimizes your inventory levels and associated costs while ensuring that your service levels are met. The reports detail what inventory levels are needed where, when, and the replenishment cycles as well as what providers move the product, when, using what modes, and at what load factor. It’s a complete supply chain plan. Furthermore, it’s easy to work with because all the reports can be output to Excel — which allows you to drill and pivot to your heart’s content until you see the data in a form that’s most convenient for you to internalize. (And while spreadsheets are not supply chain solutions — especially where optimization and analysis is concerned, they are good for report manipulation, and everyone is already comfortable with them.)

And the results are beyond what you would get with either tool on its own because not only does your distribution network dictate your inventory costs, but changes in inventory requirements over time will dictate your network costs. (If a warehouse becomes unnecessary because customer locations move and new lanes open up, that’s a considerable fixed cost that is unnecessary.) It’s a viscous cycle, and unless you look at both in unison on a regular basis, you’re missing cost reduction opportunities. Consider the case study of a major (FM)CG company in India that typically maintained about 115 tonnes of inventory in its network in an attempt to meet service levels. Not only did every tonne of inventory, depending on the SKUs in question, represent anywhere between roughly ten thousand and a few million dollars of working capital tied up in inventory, but every tonne represented additional inventory costs that chipped away at margin and profit. When Algorhythm applied their basic SKU inventory model, they were able to present the CG company with a solution that trimmed 25 tonnes of inventory out of the system without affecting service levels. (In fact, the average service level was increased!) When they moved to a multi-echelon inventory model, which balanced inventory not just at each level, but across levels (and allowed inter-level shipments as well), they were able to trim an additional 26 tonnes of inventory. But when they applied the full Strategic Distribution Network Optimization model, they were able to shave an additional 5 million tonnes. In the end, they more than halved the required amount of inventory to meet the service levels, and halved the network related costs. That’s a very considerable chunk of change that went straight to the bottom line!

Share This on Linked In

Algorhythm: Still Pounding Out the Optimization Rhythm on the Tabla (Part I)

Since I last covered Algorhythm and their supply chain optimization rhythm, they’ve been pounding out a steady beat and extending the breadth and power of their unique supply chain optimization platform. Not only do they have extensive optimization capabilities in production planning, network planning, and logistics planning — with specialized solutions for oil, steel, and packaging, but they now have a best of breed multi-echelon inventory optimization capabilities and a best of breed distribution network design optimization platform that can take multi-echelon inventory requirements into account and allow you to optimize your distribution network around your detailed inventory requirements, which can be specified at daily demand levels if you desire. This is a very powerful capability that sets their platform apart from the other solutions on the market, as most of the other supply chain optimization platforms focus on inventory, or network design, but not both simultaneously.

To understand just how powerful their new solution is, we have to start by discussing how hard it is just to optimize inventory. There’s a lot more to inventory than just the carrying cost that is recorded on the books. There’s the cost of replenishment, the cost of a stock-out, and the cost of missed service levels, for starters. If your planning is poor and you’re always having to rush inventory, or if you’re not maximizing truckload volume, you’re spending a lot more on inventory replenishment than you should be. If a stock-out results in lost sales, that’s missed revenue opportunities which go straight to the bottom line. And if you keep missing your service level targets, your customers might just find a new source of supply at contract renewal time. (And on the flip side, if you are constantly carrying too much inventory to make sure you don’t miss service levels, your carrying costs will go through the roof.)

To optimize inventory, you have to take into account the many layers of your distribution network: factories, (first tier) national warehouses, (second tier) regional / provincial warehouses, and (third tier) local warehouses; storage space at each location; valid flows from one tier to another, as well as valid flows between nearby warehouses at the same tier; transportation options available; stores or end-use facilities that require the SKUs; the individual SKU demand patterns (and [expected] forecast accuracies); lead times (and variabilities); service levels; and costs associated with storage, transportation, and stock-outs at various inventory levels. (Transportation costs in particular will vary.)

This is because you don’t need the same service level at every node in the network to achieve that service level at an end customer location, especially if a customer location can be serviced by multiple distribution centres. For example, if an end customer location can be serviced by three different distribution centres, you can achieve a 98% service level (defined in terms of SKU availability) as long as each individual distribution centre has a 75% service level (as the chance of all three distribution centres being simultaneously out of stock and unable to service the customer location is 0.25 * 0.25 * 0.25 or 1.5625%). Furthermore, as the lead time from each DC to each customer location will vary depending upon distance, transportation options, and local routes, and so on, the inventory levels at each DC can vary and still allow you to meet your target service levels, which can in fact vary by location (as you’ll want a higher service level at a high-profit location than you will at a low-profit location as service levels drive inventory which drive costs). In fact, the deeper you dive into inventory, the more complex the cost equation becomes and you see that you really do need to take into account all of the elements supported in the Algorhythm Xtra Sensory Inventory Optimizer, inventrhythm, if you truly want to optimize your inventory costs.

But this is just the beginning. Since your distribution network design will ultimately dictate your inventory costs, to truly optimize your inventory costs, you have to simultaneously optimize your network (to the extent that you are able). Algorhythm’s platform can do this, and we’ll discuss what’s involved in Part II.

Share This on Linked In

Is 2010 The Coming of Age for Sourcing and Supply Chain Optimization?

In the beginning, there was the reverse auction. Industry visionaries applied reverse auctions to their sourcing events for commodity and competitive categories (in the mid nineties) and saved a small fortune (which sometimes exceeded 30%, 50%, and even 70% of previous category costs). They were heroes and the world was good.

 

Then, a couple of years later when they circled back to the first categories and held another auction, something unexpected (to them) happened. The total savings shrunk considerably. The average savings, expressed in terms of percentages, dropped from the mid double digits to the (low) single digits. The savings often equalled what they would have expected from a traditional RFX / negotiation process. But the market was a seller’s market and the total event time, and thus the total event cost, was low, so with the right spin, they still looked quite successful. The world was still good.

 

Another couple of years passed, and they circled back to the first categories again. But this time, the market was a buyer’s market again and savings were bound to equal those seen in the initial category reverse auctions, right? Wrong! Instead, something really surprising (to them) happened — instead of saving money, total costs increased — sometimes in the double digits! The world was a dark and scary place. What happened? Could it have been avoided?

In short, as I explained in A Brief History of Optimization (published in By the Buy, the TradeExtensions Newsletter), reverse auctions are not the panacea that many auction platform providers still make them out to be and the identification of real savings through auctions can often be elusive at best. A new technology is needed, and as I have been saying (well, shouting from the rooftops) for years, that technology is optimization.

But, even though the technology is now a mature technology (as strategic sourcing decision optimization turns 10 this year, which makes it middle-aged in Internet years and a senior citizen in dog years), only the true market leaders (which generally account for 10% of the total market) have even tried it, and, in my estimates, less than half of those have truly adopted it on an organization level, even though the analysts have consistently found that strategic sourcing decision optimization consistently saves an average of 12% above and beyond what you’ll get from the best reverse auction.

Simply put, optimization is instant ROI. Guaranteed. In the absolute worst case, your allocation is already perfect and you won’t save any money. But I’ve NEVER seen this happen in practice. Even the most dismal events generally return 3% to 5% savings. Even if we’re only talking a 50M category, that’s still about 2M in savings. And now that you can run an event for (considerably less than) 100K, that’s still at least a 20X ROI!

And it doesn’t take a PhD to use it anymore. Now that most of the platforms offering true strategic sourcing decision optimization have easy to use GUIs, wizard-based constraint definition, and scenario and costing templates built right in — with full Excel integration for data collection, modification, and reporting, optimization is as easy to use as an auction platform. (And in Trade Extensions’ platform, it’s built into the auction.) And while it might still take a couple of days of training to master the advanced features, any of your senior analysts should have no problem picking it up quickly. And once they learn it, they can modify the templates for your organization and train your more junior staff, who will probably only need a couple of events to master most of what they’ll need to do on a daily basis for an average category.

And after reading this recent piece in Industry Week that says “Transformation is Out; Optimization is In” that pointed out that while organizations still want to ‘transform’ how they deliver back-office services, they typically want to move in pragmatic, incremental steps and focus on achieving best-in-class, standardized and optimized delivery models and said that while many organizations remain keen to avoid the costs of new capital and migrating to new suppliers, investment is being made in ensuring existing suppliers and internal processes are delivering optimum value, I’m starting to think that maybe optimization might finally begin to come of age. It appears that the term has finally entered the daily vocabulary of supply management professionals, who should now be more open to at least reviewing optimization solutions. And once they see the savings to be had, and the power that they can have at their fingertips, I can’t help but thinking that the followers are finally going to start to adopt this technology and become leaders in their own right. (The laggards will ignore it for years to come, but that’s okay. Most are still hunkered under their desk waiting for the recession to be over and will eventually go out of business anyway, so let’s not worry about them.)

Share This on Linked In

Another Way Supply Chain Optimization Increases Profits

A recent article in Supply Chain Brain on “Planning and Managing Demand: A Modern Supply Chain Imperative” provided a great example of how you could use scenario-based what-if optimization to slash costs and increase profit at the same time.

Another approach [to maximize the profitability of a given inventory investment] is for operations to look at the sales pipeline and see that a customer is currently in the pipeline and is expected to order 50 widgets. What if sales approached that customer with an offer of $1/widget price reduction on 20 widgets if they made a decision within two weeks? Assuming the customer accepted the offer, how much does it impact revenue and profitability?

[Let’s say that] in the original factory order, total revenue would have been $1,500 (100 at $15/widget). Cost of the factory order is $630 (90 at $7/widget) plus $90 (10 at $9/widget) for a total of $720. Margin is thus 52 percent [because a widget costs $5, a $60 container holds 30 widgets and a partial container cost $4 a widget].

[But] what is the situation if the discount offer is accepted? Total revenue for the order would be $1,500 (100 at $15/widget) plus $280 (20 at $14/widget), or $1,780. Total cost of the order would be $840 (120 at $7/widget). Margin increases to 53 percent. By cutting prices the company ends up making more money. Furthermore, it does not impact total demand (since the customer would have made the purchase anyway), but rather it affected profitability and cost, since the customer saved $20, and the company saved an equal amount.

And this is something you could easily figure out with a good scenario-based what-if decision optimization solution that allowed you to adjust prices to see what offers you could make that would benefit you and your customer(s).

Share This on Linked In