The indications of trouble are usually within a distributor's own company. It's time to take a close look at the effectiveness of your system when you hear the following:

• Customers complaining that they can't get the items they need when they need them;
  
• Financial personnel griping that inventory investment is too high, especially for non-stock;
  
• Salespeople grumbling that they're losing sales and customers due to insufficient stock;
  
• Warehouse workers grousing there isn't enough room to store all the stock; and
  
• Users gossiping about how they work around the system to take care of their favorite customers.

The absence of these comments and complaints is no guarantee that problems don't exist. Ask your customers and employees directly about the state of the inventory and listen to what they tell you.

Beyond experience

Purchasing personnel, however, should not calculate the quantity of stock to buy. Nor should they have to check each computer-generated suggested order quantity (SOQ). Their time is much more productive when they provide the human touch by reviewing computer-suggested purchase orders and using their knowledge of suppliers and customers to adjust the figures.

Purchasing people must understand the complex formulas, rules and data massaging used by modern systems to calculate suggested quantities to buy. Ideally, new purchasing employees have been educated in modern, dynamic inventory-management principles but are aware of the outmoded principles still found in too many systems.

Effective purchasing personnel should spend only a tiny fraction of their time placing purchase orders. For stock purchases, ordering should be automated, with adjusted computer-generated suggested purchase orders automatically sent to suppliers through direct faxing, electronic data interchange or integrated e-commerce. Even the purchase of many nonstock items should be heavily automated. Substantial human involvement may be required, however, to ensure that the right item is ordered at the right cost and the expected arrival date is acceptable to the customer.

Garbage in, garbage out

But even the most sophisticated systems cannot effectively manage inventory by themselves. The reports they generate also must be used in a timely and accurate manner. Just as today's purchasing people need a thorough working knowledge of inventory-related system functions, other users whose actions have an effect on the level of inventory must know both how their specific functions can influence inventory management and how other system functions can affect their own.

Some systems have formulas and rules that can't handle the erratic, complicated buying patterns of today's customers. They may not react fast enough to changing conditions or may provide imprecise calculations. Other systems use rules and formulas that are too simplistic, a weakness likely to be found in generic distribution systems, as opposed to those systems designed specifically for hard-goods distributors.

While replacing an old system is one option, changing some of the present system's rules and formulas is usually the most cost-effective way to improve inventory management. Here's a look at some calculations that may be inadequate:

• Safety stock can account for too large a portion of the quantity on hand. The old rule of using X weeks of forecasted usage for "A" items and Y weeks of forecasted usage for "B" items ignores a fundamental principle: The level of safety stock for an item should be in proportion to the volatility of its activity. After all, safety stock is kept in case sales exceed forecasted demand. In theory, the safety stock for a "C" item with a constant level of activity each month should be zero; in practice, some safety stock should be kept on hand in case of disruptions.

• Lead time frequently is calculated even less accurately than safety stock. Lead times should be based on more than just averages of several prior purchases. Even focusing on recent purchases isn't accurate enough. Again the calculation of lead time should take volatility into account, especially for manufacturers that practice just-in-time production.

  Distributors that have some form of central warehouse but also allow branches to buy direct actually have two lead times, which can be very different. Determining the right lead time requires a computer system that properly calculates and stores both lead times for each dual-sourced item and "knows" when to use the appropriate one.

• Qualifying historical data. Although most systems adjust historical data before using it to forecast future sales, many systems don't do enough. For example, consider a one-time-only spike in sales. Depending on what caused that spike, it should or should not be removed from the data or adjusted in size. The same is true for a unique dip in sales. Many systems also fail to add lost sales to actual sales. A one-rule-fits-all approach to adjusting is inadequate.

• Forecasting. Averaging prior unit sales is not an accurate method of forecasting. Averages don't take sudden up or down trends into account fast enough for buyers to react in time. They have negative effects on distributors whose products are subject to sudden, rapid changes in demand. Some systems vainly attempt to overcome this problem by using a weighted average that puts greater emphasis on more recent sales, but this doesn't do much to improve accuracy.

  Another deficiency found in some systems is the use of historical sales instead of "demand" when forecasting. This can result in not having enough stock on hand.

  If sales growth is slowing for a particular product, the better systems can adjust purchasing recommendations downward, even though recent sales have been much higher than they were only a few months earlier.

• The season that never was - and I am not referring to baseball or basketball. Although most systems calculate "seasonality," many of those calculations turn out to be wrong, especially when the "season" is based on the weather rather than a calendar. For example, a mid-spring heat wave will stimulate sales of air-conditioning equipment sooner than usual and in greater volume than normal for that period. Next year the computer's determination of seasonality will lead to buying too early and too much.

• EOQ is dead; long live EOQ. Estimated order quantity has long been inaccurate for some items, such as those that have very slow turns or very low unit value relative to the other items stocked. For these, EOQ calculates a quantity of zero for the former and a multiyear supply for the latter. Both can be handled with a manually set minimum and maximum, such as a minimum of one and a maximum of two, but it takes a lot of effort to keep min/maxes up-to-date. A better way to address this problem is to use one of several sophisticated mathematical or statistical formulas to calculate the quantity to order.

• Distribution requirements planning (DRP) vs. re-order point (ROP). DRP, another method for triggering a purchasing consideration, works better than ROP for items ordered weeks or months in advance. The multiweek or multimonth sales forecast is quite accurate. With DRP, the system determines when the safety-stock level would be reached and when a purchase order should be generated. Unfortunately, not enough computer systems contain DRP.