Stop Driving Inventory by ABC!

This article does NOT make a case against ABC analysis. Nor does it oppose segmentation strategies in general driving inventory levels. However it does explain how the common practice of combining these two is hurting your business. And it proposes alternative means to achieve the same goals with better results.

Brief primer on inventory segmentation and ABC analysis

There are many possible reasons why and possible means to separate products in a portfolio into different groups. Products frequently are grouped by the state in which inventory is held. Common are grouping by raw material vs intermediate vs finished good or parts vs sub-assemblies vs finished assemblies or some variation of these. Companies may also group product by the type of storage, such as factory stock vs central stock vs forward stocking locations or regular storage vs climate controlled vs cold vs frozen storage. The two most common reasons for such segmentation are first to gain visibility and second to drive efficiency improvement. Product companies typically manage so many products in so many different states and locations, continually varying over time it is too complex for people to keep it all under control. Inventory segmentation is the primary means to reduce complexity. If managers know a disproportionate part of inventory cost is in raw materials it allows them to target any reductions and efficiency increases in that area specifically for the greatest impact. In these cases breaking the problem into smaller parts makes them more manageable both to understand and to fix. 

ABC analysis is an extension of this segmentation concept. It allows the inventory analysis to be performed at finer levels of granularity, and gain deeper understanding of exactly where inefficiencies exist and to what degree. Rather than analyze all raw materials or all finished goods, ABC analysis breaks each such group into A items, B items and C items. A items are the most important ones, B the medium ones, and C items the least important ones. What exactly is important will differ between different companies, and how many items belong in each category differs too. Common measures of importance are revenue generated by an item or the cost of goods sold (COGS), and common splits are 5%/15%/80% when split by revenue or 20%/30%/50% when split by COGS. The highest 5% revenue generating items may bring in 60%-70% of all revenue, the next 15% may bring in 20%-25% and the biggest 80% portion of the portfolio may bring in only 10%-15% of revenue.

ABC analysis has evolved quite a bit over time. Many companies now segment into more than just 3 categories. It is commonplace to find ABCD, ABCDEF, and sometimes even up to Z. Many companies segment into more than just one dimension, creating for example ABC/XYZ matrices. In such cases the ABC dimension ranks items by some form of demand level (revenue, cost, volume, or other), whilst the XYZ dimension captures another feature of the item behavior. Common use of the XYZ dimension is a measure of demand variability, where the X items are the most stable, Y are more intermittent, and Z items are sporadic or erratic in behavior. Typical split for XYZ is 1/3 of items in each group. In this way items are now categorized as AX, AY, AZ, BX, etc. for a total of 9 possible classifications. AX is the group that contains the fastest moving (A) items that have the smoothest (X) demand pattern, CZ is the group that has the slowest (C), lumpiest (Z) demand, with all the other groups somewhere in between those two.

The ABC dimension highlights items that contribute more to the business than others, whilst the XYZ dimension separates the easy to manage from the harder to manage items.

So What's Wrong with ABC Analysis?

Actually, there is nothing wrong with ABC analysis per se. If the categories are well chosen, specific to the business needs, and it is used for visibility and analysis it is a powerful tool. But it is the wrong tool for the job when you want to manage or optimize inventory.

The practice of reusing ABC categories for inventory management started in the days when computers were not generally accessible yet and if they were present they had computing power that was a fraction of what our mobile phones have today. Inventory was managed for every item in every location by hand and later by spreadsheet. ABC segmentation became the means to determine which items should get the most of the planners' valuable time. If you knew which 5% of items were generating more than half of all revenue it would make sense to spend most of the time on those items for the greatest impact. Simple rules of thumb would be used for everything else. For example, all B items get the equivalent of 2 weeks of average demand as a safety stock level, and all C items would get 3 units each, or something of that nature. This lead to a much more manageable situation. The B and C items would practically manage themselves with an understanding that they would be inefficient to some degree, whilst the attention of the planners was focused on those items that would make the biggest impact to the business.

But times change...

Computers are now pervasive. Powerful inventory optimization systems are readily available at prices that make them accessible to even the smallest product companies. The advent of these systems has reversed the entire problem. The A and B items can be handled fully automatically by systems and it is now the C items that typically require the planners' attention. 

More importantly all this computing power allows us to no longer be dependent on simple rules of thumb. We can now drive every item like it was the most important item in the portfolio. The system can calculate optimal inventory levels for millions of item-location combinations in seconds. There is no reason not to consider every item in full detail. In fact, there is every reason to pursue it with vigor. Moving from the rules of thumb of old to approaches allowed by present day computing power will result in dramatic improvement of margins and significant, but immeasurable, increase in market share, assuming you beat the competition to it.

The two most important issues with applying ABC rules of thumb to drive inventory are:

1. The ABC classification captures only a small part of the total value of each item.
2. It treats all items within a given classification the same

With regards to the first point, let's take a step back and consider the objective of inventory management. It is to provide the highest possible service to the customer whilst providing the highest possible profit to the company. There is a trade-off between these. At low levels of service incremental cost to provide more service is low; at high levels of service, incremental cost to provide even more service may become prohibitively high. The cost-to-service curve for any item has a distinctive hockey stick shape as shown below:

Figure 1: an example of a stock-to-service curve

When you combine all items in a product portfolio or groups of items within it you can achieve similar curves spanning across all those items. Each company will decide where on the curve they want to be, based on margin, market, industry, strategy and so forth. There is no single best answer. The more service you want to provide, the more it will cost you. Some companies may even be so far right on their curve to take a loss, either temporarily or for a subset of items, to achieve some strategic objective or in support of other more profitable items.

The important thing to realize with regard to these curves is that there are many contributing factors. It is not just demand levels and variation. Inventory policies and practices, such as lot sizes, ordering intervals, and inventory review cycles, have a direct impact. Likewise supply behaviors, such as lead times and lead time variability, and supplier reliability in general, will also impact the curve for each item. Not only does the scale of the curve differ greatly by item and by location, also the shape of the curve can be vastly different. One item may be more efficient at one level of service compared to another, whilst the other may be more efficient at a different target level of service. Or one may be more efficient in one sales region whilst another is more efficient elsewhere. We cannot say ahead of time one is an A and the other a B if we consider the full impact each item has on the business. The stock-to-service curve makes the decision a bigger one. Where the ABC classification is purely inside-out with a demand perspective only, the stock-to-service curve adds the outside-in perspective (service level) and adds the inventory and supply characteristics in determining an overall efficiency to serve the customer. Some of the most revenue generating items may also be the costliest to supply or have the most unreliable suppliers causing all kinds of service level issues and their accompanying cost to expedite and loss of customers. An A item from a demand perspective may be only a B if considering the whole picture. In benchmarks we consistently see that when customers step away from their trusted version of ABC and use other groupings instead to drive inventory that service levels go up and inventory goes down simultaneously. When holding service levels steady the decrease in inventory is usually between 10% to 20%. Easy money from just grouping differently. More about that grouping further down.

Topic of the second bullet above - treating all items within a classification the same - is where the brunt of efficiency loss occurs. Not every A item is the same, not every B item is the same, not every C item is the same. Yet most companies driving inventory by ABC (or some variation of it) treat them just like that. I'll use a simplistic example, but the principle works the same regardless of what rules are used, fixed service level, fixed units, fixed coverage, or something else. Let's say a fictitious company uses ABC classification and are able to classify each item exactly right (i.e. assume issue 1 above does not apply). They decide to target inventory of A items at 1 week of demand coverage, B items at 2 weeks and C items at 4 weeks. In practice the optimal level of inventory is on some continuous scale across items, as shown in the next chart: 

Figure 2: optimal inventory level (black) vs levels by ABC class (blue)

Another way of looking at this is that in reality the lowest efficiency A item is not twice as efficient as the highest efficiency B item, rather they will be roughly the same. The cut-off between ABC categories is arbitrary. It may make sense from a reporting perspective to classify items this way, but to drive inventory by these big groups is damaging to your business. It causes simultaneous excess inventory and loss of service. The blue colored areas that fall above the black ideal inventory curve in figure 2 are items that have excess inventory. The white areas below the black ideal curve are items that do not meet the service objectives. We tend to see new customers reducing inventory by 10% to 30% whilst maintaining service levels or some lower decrease in inventory whilst achieving dramatic service level improvements. All attributed to simply not forcing all items within a single category, along whatever categorization, to follow the same rules or objectives.

When we combine this heterogeneous treatment of inventory management within groups with stepping away from using ABC categories to drive inventory levels the benefits compound.

If not ABC, then What?

If using ABC to drive inventory is damaging, then what should be the approach. Should we use different groupings or not use groupings at all? As hinted above, there is nothing wrong with grouping items when driving inventory. There is however also nothing wrong with not grouping items at all if the products are all sufficiently similar; say all consumer packaged non-alcoholic beverages, or all bulk petrochemicals.

The short answer is that the grouping should follow logically from the physical form of the product or from market strategy or from customer behavior. Treating raw materials differently from intermediates and from finished goods is a natural segmentation companies already do. This can almost always be extended in natural ways. A company could differentiate inventory of items that are highly perishable, somewhat perishable, or non-perishable; could differentiate high margin items from regular items from low/negative margin items; or finished goods from spare parts from accessories. These are just a few examples. What works for one company may not work for another. A next level of sophistication is when inventory is segmented according to market strategy. Maybe the company is introducing a new product line, and for market penetration super-high service levels are required even when it means increasing cost beyond revenue. These items would be placed in a separate group when managing inventory. If product is sold online customers might have the option to save money by subscribing to regular deliveries. For the group of such subscription items it would make sense to provide a higher level of service. Grouping could also be based on customer buying behavior. For example if an item is out of stock, how likely is the customer to buy a similar product, or a competitor's product, or wait until it is back in stock. Say you are a manufacturer of kitchen appliances. A customer for a white fridge will pick another white fridge without a thought. But a customer who wants matching units of the premium line will wait for weeks to get just the right one. You can afford to provide lower service for the group of premium products than the regular products in such cases, quite counter-intuitive. In a B-to-B situation having some products in stock even ones you lose money on may be crucial to retain customers. These strategic products could form a separate group with guaranteed high service levels where otherwise they may be targeted lower or even removed from the portfolio.

These are just some examples of how to group items. The most appropriate grouping for your business will require careful deliberation and may change over time as products traverse their lifecycles or as company strategy changes. The great thing about grouping items in such a way is that inventory will start to serve a higher purpose. It is no longer the automatic operational response to arbitrary financial drivers, rather becoming topic of strategy or executive S&OP meetings. Inventory becomes a driver of value not just of cost, becoming an important tool for the company to grow or defend market share in key areas.

Conclusion

Driving inventory by ABC in whatever variation is highly common and occurs at all levels of management. For reporting and analysis purposes it is a powerful and appropriate tool. This includes reporting and analyzing inventory performance. But when it comes to driving inventory it is essential to be more discriminatory in how to group items. Stated differently: ABC should be used on the output of inventory management or optimization, but never be used as an input.

This means not setting service levels by ABC class, or weeks coverage, or turns, or lot sizes, or even replenishment frequencies by ABC. It also means that an executive directive to reduce inventory by $X or Y% across items in one of the ABC classes does not translate to an input into inventory level setting exercise. At best it would trigger a decision on which items to retire or a strategic supply chain redesign effort. In most cases it should result in healthy cross-departmental discussion about the reason, the undesirable side-effects (lower service levels, higher transportation and expediting costs), and possible alternatives to achieve the same goals with fewer negative side effects. Easier said than done in most companies, I know. If escalating inappropriate cost-cutting directives back up the chain is not an option the only reliable alternative may be to simulate effects of various options, allowing the planners to test the impact of reduced inventory in segments of the market that are not of strategic importance to the company by cross-referencing against ABC reporting grouping.

Finally, in any inventory strategy try to refrain from applying the same rule to every item in the same group. Even though they are similar in ways, they are not identical. This is not strictly limited in use to ABC, but it seems the habit of companies using ABC to also treat every item within a classification as if they were identical. Your inventory needs an approach custom tailored to your business to maximize efficiency, but ABC itself is a one-size-fits-all approach, and then applying one-size-fits-all logic within each group causes dramatic loss of efficiency. With modern-day computing power and available inventory optimization tools there is really no more reason to keep doing this. More on this in a future article.

ABC has its uses. Driving inventory should not be one of them. 

Find all my articles by category here. Also listing outstanding articles by other authors.