Splitting the Demand/Forecast

[Former Member]

Hi All,

This one is more on the functional side. When a periodic or monthly demand (the level at which forecast is already calculated) is say split into weekly buckets- how does the supplies (fill rate) behave in both these cases assuming that there is certain level of batch/lot sizing that happens at production.

- Equally Split say. weeks demand = (Mthly demand/No of days in the mth)x 7

- Split based on the interweek sales ratio historically.

Though as planner, there were some incentives to do the later more from a rationing point of view in the event of short/non supply. but in a pure make to stock scenario inventories are in any case sufficienctly made available for shipping, i mean there is no "Pull Based"or "just in time" happening when it comes to replenishing stocks at point of sale. Shipments are made as per product availability plan and entire stocks are made available by the rqmt date (say BOW or EOW). So basically by splitting demand i am splitting the shipments as well (as per PR's created) by creating a large number of PR's with uneconomical shipment sizes. !!

or does this practice of splitting advisable from a lean organization point of view. (Pls. assume pure make to stock scenarios only). and hence helps phase the supplies evenly .. Is it desired at all.as there will be incentives to bundle the shipments which are phased..

Your opinions pls,.

Rgds/

Loknath

[Former Member · Accepted solution]

Hi Lokenath,

If it is a functional solution you are looking for, the first info required is what industry vertical you are working for? What are the products that you are planning for? What planning process do you use? etc

Generally:

1. Smaller bucket size are ideal where due date is important and the optimum lot size is small.

2. Shipment size can be optimized by using multiple batches of different products, and can be defined in the TLs.

3. Even if it is a made to stock situation, its never advisable to produce in lagre quantities.

4. The objective of low storage costs and adhearance to delivery date remains even if your shelf life is high and cost of product is low.

I am not sure if this helps but you can always take a lead from these. Also you can reply with details and Ill try to give a more definitive answer.

Regards

Santanu

[Former Member · Accepted solution]

Thanks Santanu,

All the points mentioend by you are generally true for any make to stock environment (with or without forecast consumption). I was more interested to know the service level (order/line fill rates) and inventory related costs differentials in each case.. i,e demand is split (differenet qty on different rqmt dates) and demand in lumpsum (entire qty, ome rqmt date).

Assuming that the product mix is wide enough in say CPG company (take HLL for convinience) and that all products may not be available exactly as per the net requirements calculated (i,e opening inventories are not sufficiently available for shipments by the due date(s), so some part production of the current month is also needed for replenishment), supplies in any case will be phased in the both the cases but the frequency of replenishments will be more in case the demand is split and this frequeny "may be undesirable" due to uneconomical lot sizes (+ multiple rounding to nearest shipping unit as set in MRP views) even by bundling of different products say for a end of week due date supplies to a particular location and may not result in TLs for the said destiantion. This may be particularly significant concern when the shipping cost is a significant % of product cost. This may make sense only when storage is a constraint cost wise or shelf life wise. (say refer storage)

My opinion is in real sense you don't save on the inventory related costs considering the upstream production and procurement as lot of bundling is done after clubbing the system generated production and purchase requisitions for obvious managerial reasons.

My question is over a period of time if the practice of splitting is instituted say for all products what discernable service level improvements and cost reductions can happen over non-splitting scenario. The answer may lie in empircial studies as i see. but wanted to know the <b><b>magnitude</b> of cost (inventory/storage) and service diffentials</b> in a steady state firm in splitting vs non-splitting scenarios. I buy all your points as they are valid in all scenarios and consicious planners will keep these things in mind irrespective of what the system they use. As regards "its never advisable to produce large batch sizes" - i may comment that in a large number of CPG companies and even pharmaceutcial companies often batch sizes are fixed and can't be changed, it can only be delayed or preponed. and in most case the later happens due to incentives like saving on set-ups/cleaning the lines etc esp. if the orders generated by the system are say phased by a week or so. (I would like to overstrech that ::-)).. Actually this issue arose as i was fiddling with the splitting thing in APO DP. but couldn't make out the context and why this feature is provided and as consultants who and when do we advise such a thing.

Rgds/

Loknath

Mumbai, India

[Former Member · Accepted solution]

Hi Lokenath,

A very interesting insight into demand rationalization.

My apologies for giving such generic answers as I misinterpreted your requirements.

For clarity of thought and ease of understanding we will take CPG as example in the discussion. We will try a little reverse engineering & try to arrive at a decision whether Demand splitting really helps such a scenario. Also we will discuss purely at functional level and not get into product capabilities.

Top management sets the strategic targets for brands, sales & operations. Then, senior management evaluates alternatives and works out more detailed plans how these goals will be achieved: By going into new markets, an extension of the product portfolio to new channels, by improving the existing portfolio in the traditional channels, or by acquisition or selling brands. Once the overall company plan reflects strategy, an integrated business planning process links strategic targets with tactical and operative planning on all hierarchy levels of the enterprise. All downstream plans get specific business targets, in order to ensure the adherence of strategic targets.

Business Goals & Objectives will generally be:

1. Increasing Revenue : Maximize profitability by product

2. Increasing Transparency & Accountability: Provide decision support, Increase data transparency, Improve transparency and auditability of activities

3. Managing Fixed Assets & Resources: Optimize capital equipment and asset utilization

4. Reducing Operating Costs & Increasing Efficiency: Lower communications expenses

Key Performance Indicators for measuring the same will most possibly be:

· Capital Employed Ratio

· Company Growth vs. Market Growth

· Market Share

· Net Income

· Total Planned Time

We will try to narrow theses objectives to Tactical & operational level where IT actually plays a role in the Supply Chain:

· Synchronize supply to demand -- Balance push and pull network planning processes. Replenish inventory and execute production based on actual demand.

· Sense and respond with an adaptive supply chain network -- Drive distribution, transportation, and logistics processes that are integrated with real-time planning processes.

· Provide networkwide visibility, collaboration, and analytics -- Monitor and analyze your extended supply chain.

Business Goals & Objectives in SCM parlance will be:

· Improving Customer Service

Better service levels, Collaborate with business partners, Improve forecast accuracy

· Lowering Working Capital

Reduce out-of-stock situation

· Reducing Operating Costs & Increasing Efficiency

Reduce inventory levels, Improve capacity utilization

Key Performance Indicators as defined by SCOR will be

· Demand, Supply Planning Costs: Forecasting costs + developing finished goods or end item inventory plans costs + costs for coordinating Demand,Supply process

· Perfect Order Fulfillment: Perfect fulfilled orders / Total orders * 100

· Delivery Performance to Customer Request Date: Number of orders that are fulfilled on time / Total number of orders * 100

· Field Finished Goods Inventory:

· Forecast Accuracy: (Forecast Sum - Sum of Variance) / Forecast Sum

· Percentage of customers sharing forecasts (Consensus Forecasting): Sales of customer sharing forecast / Sales * 100

· Sourcing Cycle Time:

· Capacity Utilization: Capacity load / Resource Capacity * 100

There will be many more but we will stop here. The inference from these relative to the objective of the discussions can be summarized as follows:

Forecasting as an SC activity( in terms of cost, accuracy, coordination) plays a very crucial role in achieving the business objective of a Organization.

But also Loss Sales, High inventory(leading to higher scrap or more promotions in CPG) and capacity utilization plays a vital role in revenue increase.

And Order fulfillment, Response time(agility), time to market etc. are equally important in the competitive environment of CPG industry.

Inventory costs do play a very important role over a period of time even though the cost of product may be low.

Lastly costs are inversely proportional to quality and responsiveness.

So in my opinion a lower bucket (weekly or fortnightly) profile of forecasting is always preferable over monthly or quarterly bucket, provided its accurate to a benchmark level. Also it depends on the Brand salability of the product. Unilever may not be bothered about time to market, but P&G surely will be. Forcast accuracy is equally important for both to keep the waste costs low. Whereas loss sales is important for HLL and not so much for P&G as alternative products are always ready for HLL products.

You may not agree with the approach and the explanation but we can always discuss.

Thanks & Regards

Santanu Dawn

[Former Member · Accepted solution]

Hi Santanu,

Let me first thank you for coining the appropriate term in the context - demand rationalization.. rather say manual demand rationalization. Coming down to plain common sense, I would do something like that i.e <b>splitting as much and as close</b> to historical sales as possible by taking a most conservative approach that capacity is always a constraint, and that the supply plan created (thru APO or R/3) will be accordingly <b>phased to chase</b> the split demand and that it is <b>easier to chase the split demand</b> vs. lumpsum demand under conditions of steady state. Some one has to validate that "the <b>closer the split the better the supply performance</b>- costs, fill rates and whatever KPI's are advocated to measure the supply performance.

According to me this is tantamount to doing a part of optimization that the powerful built-in algorithms are supposed to do (in case of APO). I have been seraching some papers on this topic but couldn't find any.. but still the search is on as i believe there has been enough of research (may be more than all pure sciences) in supply chain and someone would have written a Phd thesis on this topic.

I am also not convinced on the idea that splitting will improve forecast accuracy per se. as statistically there are more data points after splitting. and hence the avg % deviations of sales Vs. demand in the daily/weekly bucket will be larger than % deviation of sales vs. demand in monthly buckets (the fundamental rule of forecast accuracy).

Whats imporant here is there is an intuitive logic behind the statement that splitting the demand will result in all things good and lean KPI's that you mentioned but that needs to be proven mathematically or empirically (there has to one). I am very curious to have a look at the proof/ any observed phenomenon. This topic incidentally was debated in my earlier days as a planner too but i never was close to gettng the logic.

The purpose from an SAP consultant point of view (and spending so much time on it :- like all things sap) is that it is a very obvious question with obvious incentives and we shd be in a position to quote/unquote the necessary fundaas to the client.

Lets debate if you are in Mumbai by any chance. At this point of time i am more familiar with make to stock scenarios and may please excuse if i sound ignorant and missing something more imporant.

Rgds/

Loknath Rao

Mumbai, India

[Former Member · Accepted solution]

Hi Lokenath,

With the clarity of thought that you have on the topic I feel it will be extremly diffucult to convince my point of view.

Also I would love to continue the discussion but my schedule currently doesnot permit me to do so. Because the conversation will no longer be from the top of the mind, as it has been till now, but well thought of, logical and proven reasons are to be provided.

I am currently in Spain, responsible for a rollout with the golive date on 15th., so hope you understand my reluctance.

But I surely will catchup with you one day as I am from Pune.

Signing off as of now.

Thanks & Regards

Santanu Dawn

[Former Member · Accepted solution]

That would be good shantanu.

My present state of knowledge is a potent combination of SCM as in business and SCM as in SAP, the latter for me being just a begining. For all i understand SAP has a medicine and logic for everything. May be the asnwers lie in Supply Planning part which i haven't gone thru yet. Anyway I thank you for your patient replies. Keep posting.

Keep in touch.

Thanks

Loknath Rao

Mumbai, India