Predictable demand makes inventory management easy. Unfortunately, demand fluctuates! That fluctuation, we call demand variation. It’s vitally important to analyze demand variation so that you have a head start at attempting to predict demand. Before you can properly assess an inventory item’s demand variation, you must determine whether the item experiences seasonality. This is vital to know because inventory items that experience seasonal demand must go through a different demand variation analysis than those lacking seasonality.

What Seasonality Is
When an item’s demand is consistently higher during certain times of the year, but much lower at others, that item experiences seasonal demand variation (seasonality). For example, homeowners are much more likely to purchase leaf blowers in the fall than in the summer. Thus, leaf blowers experience seasonality. Keep in mind, that seasonality does not necessarily correlate with weather or climate patterns. If the U.S. Federal Government suddenly mandated that tax filings were due in September, then demand for tax services would shift accordingly.

What Seasonality Isn’t
We make a distinction between items that experience less demand during some periods and items that are only available at certain times of the year, such as Christmas trees. Though demand for Christmas trees, is “seasonal”, for our purposes, because they aren’t available for purchase most of the year, we classify their demand as exhibiting “infinite” variation. We’ll only be discussing items with that have some degree of demand year-round.

High Medium Low
The demand for items with seasonal demand variation does not always qualify as only “high” or “low”. Many items also experience a “medium” demand at various times of the year. That is, the item’s demand is somewhere between the seasonal highs and lows. When high and low demand is obvious, it may be tempting to focus analysis on those peaks and valleys. However, you may be able to learn even more by looking at those periods between extreme demand. It can help you determine what the “average” inventory levels should be so that you can plan for moderate demand during non-extreme periods.

Other seasonal demand items might not have much of a “medium” demand season at all. Those items may perform very well in the spring and summer, but very poorly in the fall and winter. It is necessary to notice these trends so that you can proactively plan for them.

Seasonality may be obvious for some items, but it isn’t always so clear cut. To assess the seasonality of an item, use the following process.

Seasonal Demand Assessment Steps

  1. Collect Data. For each finished good item, collect the most recent 12 months of revenue data (by month is preferred). We analyze finished goods because they are the items directly linked to consumer demand. We analyze revenue data, rather than units of measure, when analyzing many different items, to prevent us from over-focusing on trivial, low-value items. Our goal is to determine if there are clear periods of high or low demand. If not obvious with 12 months of data, look at the last 24, 36, or 48 months, as necessary, to make your determination. Never use forecast data to determine seasonality.
    While forecasts are vital for planning purposes, such as defining safety stock, they should never be used to determine whether an item actually experiences seasonality. You may have some finished goods that are seasonal and some that aren’t. While it may be a difficult undertaking, it’s worth taking the time to separate the seasonal finished goods from those finished goods with steady and consistent demand. Otherwise, looking at finished goods overall can potentially obscure the peaks and valleys of the seasonal items.
    In some cases, there is no way for you to determine seasonality before you do an analysis. Sometimes, there is just no feasible way for you to separate your seasonal items from nonseasonal items. You can still assess your seasonal demand by considering all of your finished goods in aggregate analyses. That is, you can just use math to categorize the parts and goods as seasonal and nonseasonal.
  2. Find the total revenue for all finished goods during the entire analysis period.
  3. Determine the standard workdays for each month within the analysis period. This will be easy if the schedule doesn’t increase to accommodate demand peaks. For others, a higher demand season results in six- or seven-day workweeks, and thus, more workdays per month, but these additional workdays aren’t necessarily standard. We only want to account for standard (“regular” workdays. Otherwise, when we calculate average daily revenue for each month, there’s a risk of daily revenue appearing “normal” or even “low” during peak season.
  4. For each month, calculate the average, maximum, and minimum daily revenue (Table 1). To calculate the average daily revenue, divide the item’s total revenue for that month by the month’s total standard workdays. We use workdays rather than calendar days to accurately assess demand in months with much higher or lower workdays. Months filled with holidays, such as December, might have as few as 15 while others, such as March, may have as many as 23. In this case, while each month has 31 calendar days, not all of those days have an equal opportunity to impact demand.
    Assessing Seasonal Demand

    Table 1. Monthly Demand Metrics

  5. For every item, calculate each month’s demand (average daily revenue) as a percentage of annual average daily revenue.
    Month Demand % of Annual Demand = Month Daily Average Daily Revenue / Annual Average Daily Revenue
  6. Categorize each month’s demand level as either low, medium, or high. Use the following guidelines to get started (adjust as necessary):
    Low = Month Avg DD % of Annual Avg DD <= 80%
    Medium = Month Avg DD % of Annual Avg DD > 80% < 120% = ~100%
    High= Month Avg DD % of Annual Avg DD >= 120%
    Note those cases where the percent of annual revenue shifts from low to high (or high to low) from one month to the next without a medium period in between. This means the change is significant and sudden, rather than gradual. You will need to plan ahead for these demand shifts to avoid costly delivery or overstock issues.
  7. Calculate the average daily revenue for each seasonal category.
    Category Daily Demand = Total Demand of all Category Months / Total Workdays of all Category Months
  8. Identify items with significant seasonal demand variation. A good rule is to define those items with at least a 30% demand difference between any 2 demand period categories as experiencing significant seasonal variation. That is, if high daily demand is 30% or more of low daily demand, then the item experiences significant variation.
    These items should be flagged, and their demand variation should be analyzed in distinct seasonal demand periods, rather than in annual aggregate.
  9. We determine seasonality by assessing finished goods. But, remember that all items that have seasonality will need to be replenished on an individual basis. Since the items with seasonal demand variation will need to be resized any time there is a seasonal shift, they will require a multi-card kanban solution. Be sure that your kanban calculator or spreadsheet has a way to account for seasonal items.
  10. Determine the lowest annual daily demand for each seasonal item. This is necessary so that you can determine the proper kanban order quantity (KOQ) to deploy during the low season, using two-cards.

Daily Demand and Seasonality
The daily demand is sure to change from season to season for seasonal items. Of course, the period of high demand will result in higher averages, highs, and lows for daily demand than the period of low demand. Because of the difference, resizing is necessary to ensure you have the right kanban solution. While you may be able to predict some forecast values,

Items that experience seasonality, require multi-card kanban solutions. The significant demand correlating with seasonality dictates that the number of kanban cards deployed in the multi-card kanban solution is regularly re-calculated. When performing this calculation, don’t forget, to always use current daily demand to determine the appropriate number of kanban cards.

  • Current Daily Demand = (actual usage right now): Used for kanban resizing and determining safety stock.
  • Low Daily Demand = (daily usage during low season): Used to define kanban order quantity (KOQ).
  • Average Daily Demand = (for year or medium demand period): Used in demand analysis, as a reference.
  • High Daily Demand = (daily usage during peak season): Used to calculate an item’s maximum number of kanban cards.

Summary
Properly assessing demand variation is vital to inventory management success, starting with defining proper safety stock and inventory levels. However, before you can assess and item’s demand variation, you must determine whether it experiences seasonality.

Seasonality should be assessed at the finished goods level. If an item’s average daily demand during any period is 20% or higher than what it is during any other period, then the item experiences seasonality. Seasonal items require multi-card kanban solutions.