Facing Seasonality – the true problematic issue

Peak in second season

A common paradigm is that the only line of action required to face seasonality is updating the size of many buffers according to a forecast of the demand growth. Suppose the target level of SKU1 is 100, and during the season the demand is expected to go up by 100%, then the simple action is to increase the buffer to 200.

REALLY?

Does this buffer increase ensure excellent availability of SKU1 during the season?

The main flaw in the above paradigm is the assumption that the replenishment time remains the same within the season.  In too many cases this is simply wrong.

What impacts replenishment time? We know many flawed policies that stretch the replenishment time, but when we push all those policies aside, we still have one variable with very considerable impact: the amount of excess capacity on the weakest link.  We know that when the actual load approaches 100% of the available capacity of just one link then replenishment time goes up to the sky.  Try simple simulations to demo that effect.

Can the distribution channel put all the responsibility to quickly replenish the supply during the season on the shoulders of the suppliers? Of course they can, the distribution channel is usually the big gorilla that dictates the business rules.  However, what happens when that does not truly work?

Before we proceed let’s check what are the characteristics of a “season” that could impact the solution. I like to differentiate between two very different “seasons”.

One is a significant peak of demand for very short time, due to a holiday or a public event.  Short time means it is shorter than the replenishment time.  Without the ability to replenish the TOC basic approach is not usable.

The second type is a long peak allowing enough replenishment orders, thus providing excellent availability without maintaining too much stock.

This is the meaning of a “season” for this post.

How should we know the replenishment time during the season?

My argument is that we cannot know a-priori, given that the load on the weakest link, also on the next most loaded resources, is assumed to go up significantly.  In reality we cannot get a reliable mathematical function that predicts well enough the lead-time, given the realistic dependencies between loaded resources, and the internal politics within the organization that make an impact on the behavior.

What can be done is taking actions to provide enough protective capacity in the shop-floor, even within the season, to keep the replenishment time close to the time before the season. There are two ways to reduce the load during the season:

  1. Reduce the product-mix during the season. This move reduces the number of setups, and it also requires less total inventory during the season.
  2. Preparing high stock of several fast movers to potentially cover all the demand throughout the season.

My observation is that the majority of the organizations offer too wide product-mix. The topic of too high variety has been dealt with in a previous post.  The point here is that the distribution channels should strive to reduce the offered variety at least during the season, where the damage of overall too high inventories, while suffering from serious shortages, is pretty clear.

The second way is much more drastic than just increasing the buffer. The objective is to free capacity in order to be able to quickly replenish all the rest of the product mix.  The choice of SKUs for the stock required to free capacity has to consider parameters like relatively low level fluctuations and also good enough demand after the season, so remaining stock could still be sold.  Usually fast movers apply better to these criteria.

As already mentioned the distribution channel should consider the problem of lack of capacity at the suppliers as relevant to its own Strategy. Certainly any change in the product-mix requires direct dialogue between the supplier and the channel. When the supplier might still lack capacity to ensure good availability throughout the season the channel should support the supplier to face the risk in producing very high stock of several items, for instance, by a commitment to a certain quantity to be sold within the season.

Producing stock to be used during the season does NOT mean the target level should be that high! When the extra stock is for freeing capacity then the inventory is intentionally much above the target level.  Only when the total inventory in the system goes below the target level replenishment is required.

Thus, in preparation for the season two different decisions have to be taken:

  1. The new target level required to support the higher level of sales at about the same replenishment time.
  2. Choosing few items and the quantities to be produced before the season. The quantity should be based, at the very least, on the pessimistic forecast of sales within the season. Overall the capacity to be freed should allow enough protective capacity to keep the replenishment time intact.

Another insight to remember: reducing the target levels to their original size one replenishment time before the expected end of the season!

You certainly don’t like to replenish just before the demand goes down. Note, DBM is far too slow to note the start and end of a season.  The prediction of a change in the sales trend should be handled by intuitive forecasting.  Mathematical forecasting of seasonality requires data from several years back, which is possible only when the key parameters of the market have not changed during those years.

Promotions are self-inflicted seasonality. Their durations are usually shorter than a season and longer than a peak based on an event.  A dedicated post on promotion is due in the near future.

Advertisements

Published by

Eli Schragenheim

My love for challenges makes my life interesting. I'm concerned when I see organizations ignore uncertainty and I cannot understand people blindly following their leader.

3 thoughts on “Facing Seasonality – the true problematic issue”

  1. Lovely summary, as always, Eli! Most people forget the impact of increased demand on Replenishment Time and variation.

    The only thing I think people should be aware of is that when seasonal demand rolls up or down, as opposed to when the start or end of the season’s demand is more like climbing or dropping off a cliff AND Replenishment Times are short, relative to the pre-season or post-season demand increase or decline, DBM may be fast enough to adjust Buffers without manual attention. Such short replenishment times are what managers should be trying to realize, as Eli wrote above.

    So while I generally agree with “Note, DBM is far too slow to note the start and end of a season.” I’d add “usually” between “is” and “far”

    Like

  2. Great post Eli, especially when many companies make most of their revenue during these seasons. I would like to add another problematic issue regarding the new target level required to support the higher level of sales.

    Not only do we have to establish this new buffer level, but we also have to decide in how many steps and when should we start building these buffers and producing the stock for the SKUs we decided to produce before the season. The answer depends on how much excess capacity we have available, which in turn requires aggregated forecasts of regular consumption.

    In our experience, dealing with seasonality, especially in retail companies with many stores and thousands of SKUs is a real challenge!

    Like

  3. Good text Schragenheim!

    I and my teacher just finished a research of MTA. We simulated a flowshop line. In one of the scenarios demand was seasonal. We can show that you write in the text.

    I would like to invite you to participate in the writing of the article with the research results.

    Feel free to accept or not.

    You can responsing by e-mail in robson.afl@gmail.com.

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s