Is it appropriate to add safety stock to forecasted finished goods demand? In different words, should safety or buffer stock be included in the forecasted demand for products, or be driven separately in the planning system? From my perspective, the basic difference between adding a buffer quantity to the forecast versus driving safety stock in the planning system is that a buffer quantity in the forecast could be consumed by actual customer orders if the buffer is realized by actual customer orders. On the other hand, safety stock (depending on the approach utilized) will plan to keep a certain level of inventory regardless of customer order levels. Conceptually, I am wondering what the best approach is for this.
Do you have any experience or insights into this? Please let me know your thoughts.
My organization is debating this same topic as well. I only think safety stock is of value when it is used to smooth out demand or if you have a supplier that cannot meet demands on time. But ss belongs at the component levels for that reason, if you put it into the finished goods then everything on the boms will be driven to excess inventory over and above the normal forecast. The ERP won't know the difference and anyone buying or planning from their order action make or buy reports will overplan and/or overbuy. We may be going with taking ownership of the forecast from the supply chain standpoint and ensuring that it is updated regularly or at least reviewed for unplanned demands. SS is only going to increase your overall inventory on those components that you can easily get if it is applied at the FG level. You would need to compensate that process with level of ROP or better yet a kan ban system for most parts.
Good question Max. Part of the answer will depend on your industry. In CPG you have to weigh the cost of safety stock versus an empty shelf when a buyer wants to purchase. In pharmaceutical, on time is king so they may have buffer inventories but have to balance that with expiry. Generally speaking though I agree with Kevin, safety stock belongs at the component level or at the lowest possible level of the BOM. We have seen some customers develop postponement strategies, set safety stock only on critical, long lead time items (no need to use safety for those easy to get items) and use dynamic safety policies that re-adjust based on increase and decrease in demand. A solid review of safety stock policies that continually balance inventory against delivery can drive value for both customer and supplier. Kevin also makes a good point on the use of ROP and Kanban, two very effective strategies for reducing inventory and lead time. I can take from this it’s not a one size or one strategy fits all. Knowing the different strategies and applying the right one in the right place will be key to your overall safety stock approach
Doesn't it depend on the level of postponement? If you are promising from FG, isn't that where you need to keep the SS? This is also related to your supply chain agility and cost structure, which have a direct effect on your MOQ. This is the very reason that postponement strategies are so attractive, but also why an agile supply chain is so important if you are goign to use postponement. Unfortunately you have to keep inventory/SS at the elvel on your supply chain at which your order-to-delivery lead time equates to the customers expectations. If the custoemr expectation is only 1 day, then it is very unlikely you can get away with anything other than minor kitting.
Would a dynamic SS e more sensible? If you use a periods-of-cover measure, instead of a fixed quantity, then your SS will vary as the forecast goes up or down.
Great point that you make about supply chain aligity and your ability to respond and I agree with you on that. Just because you have a safety stock built into the FG doesn't guarantee that you will have product on the shelf when you need it. All s/s does is add a buffer that drives additional demand in the supply chain so that when an unplanned demand comes in, or a rejected FG, or any other type of breakdown that is not predicatable does it really come into play. You said it the way I see it as well, it depends upon your cost structure, your supply agility, and how fast you can respond to a demand when something goes out of the normal bell curve on a forecast or some other disruption. The problem at least for me is that could be everything so do you apply s/s everywhere? Can you afford to? Do you have the space for all that additional inventory? When do you reduce the levels of s/s? Everyone has a super stock list or the never can be out of stock list and every company calls it something different, I can see where safety stock on those FG's may be of value but I would then argue that if you identify the bottlenecks in the supply chain whether they be internal or external and focus on a corrective measure to compensate or fix that then you can reduce s/s levels or eliminate them and move the inventory further down the supply chain to ensure that the supplier who can never deliver on time or the process that is too complex like "plating", or a housing coming in from China, or there is never enough labor in a certain cell that is where s/s can be of value. You are correct, whatever the timeframe it takes you to replenish your inventory based on what sales and marketing is telling you they need in stock in order to meet the customer demands is where the level is set at. However, you can also manage the forecast, or do something else we used to do and that was to create "planning bills" that used common components but could be made into almost any FG configuration within a set period of time. That way we enjoyed the benefits of reduced inventory but had the ability to build sub assemblies to a certain point but not finish them until the actual order came in.
I saw a post from Bill just before and he said it the best way, there is really no one solution, it is really a combination of activities all within the supply chain that will ultimately work the best and even then we have to constantly manage demand because of the volatility of every industry. That is another reason why people in Sales and Marketing want more inventory on the shelf, but it is the Supply Chain group that is held accountable for both service levels and inventory levels, that balance is the key and is very difficult to achieve as I am sure all of us can attest to.
Thanks for your time and this is a great thread, I am really enjoying reading all the response.
This discussion would depend on the industry and type of product. I agree with Kevin that SS is best at lower levels of the BOM. My former company was assemble to order and they intentionally inflated the master schedule with a buffer of options to cover the uncertainty of what customers might order. In that case orders should consume the forecast. The MPS was continually reviewed and the buffers adjusted to insure that extra material was driven by MRP.
We looked at inventory optimization systems which compute safety stocks at lower levels of the BOM but a significant stumbling block with those is they obscure to the master scheduler what options they could promise. If the MPS has a forecast for an option it is easy to know you can promise it but if you have to run a full MRP explosion simulation to find out what you have, that gets very complex and time consuming. In this industry it was worth the inventory investment to insure that a missing option would not delay a customer's order. Having a buffer at the MPS level made it very visible what could be promised.
Then there is the debate about whether or not you want people to expedite to fill safety stock or not...
This certainly is a very good question, and goes to the basics of the planning system.
Typically, you need finished goods safety stock to cover the variations in demand, which can never
be avoided in actual practice. The decision is also not related to the type of industry or product.
The type of product (pharmaceutical or other) only influences how much safety stock you plan for.
In other words, if for one moment you assume that you can forecast demand 100 % accurately,
then you need no safety stock at all at the finished goods level, no matter what type of industry.
This also assumes that you can also produce the item with 100% reliability.
The answer seems to be that you need safety stock both at finished level and component level.
Both of them are dependent on the variability of usage or demand and on the lead time.
This is a great question especially when you consider safety stock has been a subject of countless strategies regarding service vs. investment and cost. One fact remains through all of these ageless debates is that safety stock is a static inventory that can only be changed or removed by resetting the static value placed in a planning system. This has led to even more numerous studies on how often to reset safety stock levels and what method to use to set them. Two pitfalls are the biggest dangers of safety stock from my experience; 1) MRP systems will always create demand over and above planned or actual demand using the safety stock level so that the inventory at all points upstream through the supply chain will be required to satisfy the added requirement, and 2) the more often this level is adjusted the more the planning system creates a bullwhip effect upstream into the supply base. Incorporating safety stock into the planning volumes therefore poses a great point. Again my experience is to do this in a non conventional way which sets reorder points to a fairly steady state level of consumption, this represents traditional forecast volumes, and incorporate a flexibility factor based on demand volatility, this is analagous to safety stock except it is dynamically integrated into the planning system. Combining this technique to set ROP levels and let the supply chain run to a replensihment scheme takes you to a strategy far removed from the safety stock terminology at least in the traditional sense.
THe bottom line is that combining safety stock into the planning system is a path to a pull system demand based management process that can take forecast execution and traditional safety stock routines out of the equeation including the errors that come with these older techniques, i.e. forecast accuracy and padded inventory management driving overstocks, stock outs and expeditied logistics.
Even with an ROP system that will lead to a demand pull system aren't we always implementing some level of safety stock into the equation to account for demand volatility regardless of what method we are using to drive demand? Every time we go to analyze it always comes up that we have to add in something to buffer the replenishment because something inherently doesn’t go right and then we miss a deliver. To compensate for that there has to be some level of safety stock in the equation whether it be at the component level or the FG level, right?
I agree with you. I would rather keep my forecast 'clean'. Undoubtedly this is an issue with ERP planning systems as Jeff states, but I fear that adding addition demand to the forecast is a cure that is worse than the disease.
I would much rather companies implement a 'periods of cover' inventory policy. The policy may be too keep 3 weeks of demand in stock because that is the supply lead time to the stocking point. If the forecasted weekly demand is 10, 20, 20, 30, 10, 10 then the SS would be 50, 70, 60, 50 over the next 4 week. In other words the safety stock fluctuates with the anticipated demand.
I would not disagree with either statements above, yes we do want to keep the forecast clean and yes we do want to have the inventory safety factored at the various levels of FGI and components, how much at each inventory area depends on consumption lead time visibility vs response to replenish capability. There is a baseline steady state that needs to be set for a demand plan based on demand volume that has been serviced in the past, recent trends and projected factors that may affect trends of the steady state. This will be the clean part of the forecast as Trevor states and will usually cover 80% of all demand over a given period. This is analogous to even the best forecasts being no more than 80% accurate.
Then there is the variability portion of the demand that covers the other 20% of demand over the period. This has to do with events in the market, some are predictable some are not. As steady state changes there will be a proportional increase or decrease to the flexible reserve. Setting ROP can then be done based on the 80% baseline, and the maximum can be set at the baseline plus flexible reserve. Replenishment is set based on the 80% baseline until we fall below the ROP min point then replenishment adds the flex reserve. On the other side of the replenishment we will stop replenishment for a period anytime we achieve the maximum level. Trends are adjusted through sub periods and replenishment periods make up the smallest sub period. example: planning periods for analyzing baseline, flex and associated ROP levels are months, this is where the 80% and 20% may also adjust based on trends vs. volatility. Adjusting the baseline, flex and ROP quantities are on the order of weeks, and adjusting replenishment quantites is on the order of days. Periods are aligned to replenishment cycles which synchronizes the supply chain execution and logistics.
In this environment forecast is only one element of the demand management and execution with inventory consumption (some of this is allocation) being the larger factor. Safety stock becomes a flexible reserve that has it's unique set of rules for replenishing and adjusting, these rules must be aligned to demand trends, i.e. the baseline and market volatility factors to provide the best service with minimal stockouts.
Bottom line is that trying to put SS in traditional MRP forecast planning only gets messy, to Trevor's point keep the forcast clean. To put SS in a planning routine requires changing methods toward integration of inventory sizing based on demand planning, demand management and a replenishment linkage to ROP (Min/Max) levels as a primary feedback mechanism.
This is a good discussion that has perplexed supply and value chain people for years, especially those living it daily.
I will post more info on this to http:/bpcg.net if you're interested.
Your last statement is absolutely correct and I had to read your post twice to really understand what you were saying. I agree with all of it although given all the contingencies that we use to manipulate ss, when your boss or whomever comes to us and a corporate objective is to reduce overall inventory isn't safety stock one of the first places that we look to cut? When we evaluated safety stock in dollars, we came out to 1.7 million dollars and by "peeling back the proverbial onion" it lead us to work cross functionally with sales, marketing and mfg to take safety stock levels down to approximately 300K and service did suffer by about 8 points (90's down to the 80's) but overall it was worth the effort. I am curious if anyone has come across this and attacked safety stock in terms of reducing and/or eliminating for this reason.
Using periods of cover is a great idea but what happens when corporate mandates inventory reduction, no drop in service levels, and tells supply chain that they now own the forecast and can now fix demand planning since Marketing certainly cannot do it. I still think safety stock is the first casualty of this process. Have you had experiences similar to this?
My thoughts are as follows:
Purpose or Goal of Safety Stock: To protect against supply delays/Disruptions there can be several other reasons
Purpose of Forecast : To get intended sales in the form of foreacst- Demand.
Analysis: Safety stock needs may be varying depending on Product lifecycle and the Market seasonality/working capital etc.
By keeping it seperate it allows you good control to identify /peg it. Plus Working Capital needs can be seperated for better analysis and supplier performance. So Best Practice will be to keep it seperate unless there are other strong reasons to combine it with forecast