Not long ago, I attended an industry roundtable discussion where a number of senior logistics people involved in the international movement of goods were discussing key performance indicators (KPIs). These people represented a range of industries and segments of the logistics chain, including shippers, freight forwarders, and owners of assets used in the movement of cargo. I was surprised to find that although "key performance indicator" is a basic business term used by almost all companies, there was little agreement on a definition of KPIs or on which ones should be used in global logistics.
This was troubling, because without this basic understanding, coordinated work and focused improvements across service providers and cargo owners in the logistics chain (the flow of goods from the source to the end customer, along with the relevant information and payments) are not feasible. The average international move involves around 25 entities, 35 documents, and multiple legal requirements, including applicable laws in the sending country, the receiving country, and for the modes of transportation (land, air, and ocean). An international shipment can also potentially involve multiple financial standards and accounting practices. Adding to this complexity are the different company and national cultures that influence business decisions.
![[Figure 1] Normal curve with different standard deviations](/media-library/20160302simple-kpis-global-shipments-ex1.jpg?id=53611941)
KPIs are necessary to ensure that all of the many parties involved are focused on the same objectives. Otherwise, each of the 25 entities would focus on whatever is most important or provides the most value to itself. This local optimization inherently results in an uncoordinated, inefficient logistics chain.
To help companies create coordinated, efficient, and effective logistics chains and supply chains (the logistics chain plus the process of procuring the goods through purchase or manufacture), this article offers a definition for key performance indicators, lays out what makes a good KPI, and suggests KPIs that together provide the information needed to properly manage any global logistics move.
What makes a good KPI?
A common mistake is to believe that the terms "KPI" and "metric" are interchangeable. They are not. While all KPIs are metrics, not all metrics are KPIs. A metric is any measurement of a specific part of a supply chain operation. It can be as simple as the time of departure of a transport or the correct completion of a document.
A KPI, by contrast, is a metric that is derived from and directly supports the business objectives of the company. One example of a business objective in a logistics environment might be to add value to the organization by offering the most cost-efficient and service-effective logistics capability. The organization could strive to achieve that strategic objective by meeting related, tactical objectives, such as having the ability to deliver goods on time in any country, achieve the lowest costs for the service being offered, and solve problems before they become major issues or the customer becomes aware of them. As suggested by its very name, the KPI indicates how well the organization is performing against these and other types of business objectives.
Ideally, the various KPIs should be expressed in economic terms so there is a common base of measurement. This helps to ensure consistency and a clear focus by all the participants on the same performance criteria.
KPIs must be applicable to each entity as well as to the entire logistics chain. When added together mathematically, the measurements for each entity create the overall KPI measurement. Valuable KPIs will be simple measurements, preferably derived from operating data that is available as soon as the goods have a proof of delivery for the relevant segment of the move as well as for the final delivery. Waiting 45 days for accounting data to be updated is not ideal. There must also be sufficient data to make the measurements statistically reliable. For this reason, transportation lanes with very low volumes may not be statistically able to support all the KPIs proposed, and therefore may have fewer KPIs.1
And finally, for a KPI to be effective it must help all the parties in the supply chain to work toward the agreed-upon business objectives. The KPIs selected should benefit all parties involved, not just the party paying for the move. Moreover, KPIs should make each of the entities involved in the logistics chain aware of what the overall measurement is and how they contribute to the success (or failure) of the entire chain.
KPIs for global logistics
In short, KPIs for a global logistics chain should fulfill the following criteria:
- Focus all the parties on the overall business objective;
- Be measurable both for the overall supply chain and for each of the parties involved;
- Be expressed in a common base of measurement that can be translated to economic terms; and
- Be easily accessible and statistically reliable.
While identifying KPIs for a global logistics chain, it can be helpful to remember that each international logistics move is a process that starts at the source and ends at the proof of delivery. This process involves three different movements: the physical flow of the goods themselves, the attendant flow of information about those goods, and the flow of payments (for example, the payment of duties) to ensure the goods move expeditiously. KPIs for global logistics should focus attention on and enable the management of these three different flows.
KPIs for global logistics should also measure the value the logistics process can add to the business objectives. Typically this involves delivering the right products to the right place, at the right cost, and with the right service standard. (The service standard establishes what constitutes on-time delivery, as well as the allowable failure rate; for example, a service standard of 99.8 percent on-time delivery allows for just .2 percent of deliveries to be late).
One KPI that is recommended in the literature is the Economic Value Added (EVA) from each entity for the movement. EVA is calculated by taking the revenue, expense, and capital involved and, using a somewhat complicated method, adding them into a single measure. This is a very good way to measure the value added by each entity to the process in economic terms.2 Having this information also would be helpful in ensuring that each participant in the logistics chain receives an appropriate share of the profit. However, companies are very reluctant or unwilling to share their financial information, making it very nearly impossible to apply EVA in practice. This is particularly true for a global logistics movement. In an international logistics chain with 25 entities and multiple legal, financial, and accounting standards, it is highly unlikely that the party controlling the logistics chain will be able to get the EVA information it needs from every one of those entities. Therefore, the measurements used in a KPI have to be practical, operational measures known to each entity, rather than detailed financial measurements from each one.
For example, as noted earlier, every global logistics move occurs along a particular lane (defined as a route between a shipment's origination point and its final destination). This move stretches from source to proof of delivery and is made up of individual steps or processes that are required for successfully carrying out the end-to-end move. These smaller processes often include adaptations for the specific lane, such as generating a customs document that is unique to the destination country. Each lane has a cost (the dollar or other currency value of payments for logistics-related activities, such as transportation, warehousing, and freight movement, among others), and every process has a cycle time with its standard deviation (which measures the amount of variation from the mean). When applied together, these measures enable the complete control of the logistics process.
These criteria can be measured for each entity involved in the process and can be agglomerated statistically into a total logistics chain value. As the cost, cycle time, or standard deviation decreases, the logistics chain becomes more efficient and effective and directly supports company objectives. Additionally, as will be discussed in the following sections, those measures can be converted to economic values, allowing the creation of a composite index in economic terms, rather than simply a series of statistical values.
Cycle time and working capital
Cycle time, sometimes called lead time, is an operational metric that can be measured both for the entire logistics chain and for each entity in the chain. In logistics, it begins when the goods are available from the supplier and ends when the buyer pays for the products and services. Every logistics chain has a practical, average cycle time, and every logistics chain can measure the standard deviation from that average.
Cycle time for some processes within the logistics chain, such as discharge from a vessel and final delivery, offer opportunities for improvement. For others, significant improvement is unlikely, as, for example, ships have a maximum speed and cannot sail faster. This is why it is important to measure the cycle time and standard deviation for each process step in the logistics chain; without that information, it would be impossible to correctly identify the causes of delays and unreliability.
It's difficult to achieve the average cycle time on a consistent basis. That's because the cycle times for the various processes in any logistics chain will be variable. It's likely that delays will occur due, for instance, to inaccuracies in information, delays in payments, or customs holdups, to name just a few possibilities. As a result, the distribution curve for actual cycle times will not be evenly balanced like the curve for an average cycle time, but is skewed toward later deliveries (called "positive skewness").
Accordingly, simply measuring average cycle time will not give a good enough sense of how well the logistics chain is fulfilling the business objectives or how "in control" the logistics process is. The average only tells us the best estimate of the cycle time, but it gives no indication of how reliable the delivery process actually is. To measure this it is also necessary to know what the standard deviation is from this average cycle time. The standard deviation, symbolized by the Greek letter sigma (σ), quantifies the amount of variance or dispersion of a set of data from its mean. A low standard deviation indicates that the data points are close to the mean, and a high standard deviation indicates that the data points are spread out over a wider range of values.
For example, a logistics chain with an average cycle time of 100 days plus or minus 7 days is far more reliable than a logistics chain with an average cycle time of 100 days plus or minus 70 days. In Figure 1, all the curves have the same average, and for simplicity are symmetrical, but they range from very well controlled (blue) to poorly controlled (green) processes. The standard deviation σ in the legend shows that a higher standard deviation equates to a less well-controlled process.
Cycle time measures goods in transit. These goods have an economic value during this cycle time, which is the working capital that is used to procure or is committed to the order multiplied by the value of the capital (the rate for using the working capital). Cycle time, therefore, can also be expressed as the economic term "working capital involved." The longer the cycle time, the more working capital is tied up in the movement of goods.
It's important to note, however, that working capital is affected not only by the time period involved in a cargo movement, but also by the increased inventory that an unreliable service will require in the country of distribution. The length of the cycle time directly affects how much inventory there must be in the receiving country to compensate for variations in consumption or delays in shipping. While the inventory can be calculated, in order to simplify the process we use 2 σ, which is greater than the average cycle time, to represent this added inventory carrying cost.
When calculating the working capital in a logistics chain, we do not need complex equations, such as those used to measure skewness. Instead, we are looking for a value that approximates the working capital without excessive or detailed measurements and also incorporates the cycle time and the variance. A decline in these two measures of physical distribution performance indicates reductions in working capital.
It is recommended that the measurement for working capital involved be equal to cycle time plus 2 σ (in days). We choose 2 σ because it includes 98 percent of the deliveries, and thus we only have a limited number of excessively late deliveries.
We can further translate this measurement into an economic value. The working capital rate for a company is measured using its "weighted average cost of capital" (WACC), a measure of a company's cost of capital in which each category of capital is proportionately weighted. This measure is around 12 percent for most companies and can be used for the whole logistics chain.
So the economic value of the working capital involved in a global logistics shipment is:
WACC * Cost of Goods Moved per Year for the Lane * (Cycle Time + 2 σ as Days) / 365 days
For example, if a shipment's cost averages $100,000, and the shipment takes 25 days (20 days for the cycle time and five days for the variation of 2 σ) to be delivered, then the economic value for the working capital per shipment = 12% * $100,000 * 25 days / 365 days = $822. Of course, if there are 52 shipments sent to a location each year, the total per annum would be 52 times this number, or $42,739.
Excessively late deliveries
Another thing that all logistics chains should measure is service reliability. It can be a nightmare for a logistics organization if even one box of product is delayed or temporarily lost. There is the potential, for example, that the missing or late box will contain the item that delays an entire job. Contracts often contain lateness penalties, and even if there are none, a lack of reliability will have consequences for future contracts. The consequence of this kind of exception is different in each industry. If a shipment of consumer products is lost or delayed, it may cause some customer dissatisfaction, but the penalties will be less than if an industrial product is lost or delayed and production or a project is stalled.
We need to find a practical way to measure exceptions that can have major consequences for a logistics chain's reputation. We could apply theory around order quantity and safety stock, but that is theoretical and complex. As noted earlier, 2 σ means that statistically we have delivered 98 percent of the goods on time. Taking a practical view, the occurrence of any delivery that exceeds 2 σ is a problem. An adequate metric, then, would be to measure the exceptions to the cycle time above 2 σ and apply a penalty factor, which is:
Average Number of Containers Excessively Late per Month (>2 σ) * Average Revenue per Container (of All Moved Goods)
The calculation of the number of containers that are excessively late must be an average over a specified period. For example, one could use the current month plus the previous two months. This rolling three-month calculation would be updated each month. (Containers are used here only as an example. If breakbulk is being moved, then an item or box can be substituted.)
Creating one number
We now have three economic KPIs that can be used to measure global logistics moves: the cost of the lane, working capital involved, and a penalty for excessive lateness. Each of these can be calculated by the individual entities in the logistics chain using the costs, the cycle time, and the standard deviation plus the WACC, the cost of goods moved per year, and the average cost of revenue per container.
While the company that is paying for the move (which could be the shipper, the buyer, or some other entity) may not wish to give exact figures for the cost of revenue per container and the cost of goods moved, that should not be a problem, as a KPI can be effective even with nominal numbers that do not vary. The revenue figure used to indicate late delivery could be as simple as two times the cost of the goods. So, for example, if the company does not want to assign a revenue value to a container, it could assign a simple factor—say, 2x—to the value of the container load. This can be used to 1) disguise the revenue, and 2) show the value of the delayed or lost cargo. This factor should change depending on the type of cargo to reflect the revenue lost or the penalty factor for late deliveries. For example, the factor for high-value industrial products would be higher than for consumer goods. The important thing is that the numbers give credence to the value calculations, focusing the service providers on the size of their contribution without giving away true revenue.
Improvements in these three KPIs will represent improvements in the logistics process. In other words, as the lane cost, cycle time, or standard deviation decreases, working capital and the risk of an excessively late delivery also decrease.
If desired, these three KPIs can be consolidated into a single index that will allow the measurement of the total performance of the logistics chain. The index would be calculated as:
Index = Cost of the Lane + Working Capital + Excessive Lateness Penalty
Obviously, the goal would be to make this index number smaller. One could start by establishing a baseline index number and then applying the calculations to show the extent of improvement (hopefully). So an even more valuable performance index would be:
Index = {Cost of Lane + Working Capital + Excessive Lateness Penalty} Initial Values / {Cost of Lane + Working Capital + Excessive Lateness Penalty} Current Values
The larger this index value, the better the performance of the logistics chain.
Using KPIs to identify problems
Once the KPIs discussed above have been standardized across the entire global logistics chain, they can be used to identify problem areas. Here are just two real-life examples in which these KPIs were used to identify problems and drive improvements. The first shows how using KPIs can help companies determine the source and validity of costs, and the second involved using KPIs to identify problems with cycle time.
1. A single freight forwarder handled all aspects of a movement from the source to the destination country warehouse, including the inbound customs brokerage and transport. The KPIs discussed in this article uncovered performance problems that made it necessary to replace the inbound customs broker, which was owned by the freight forwarder. The forwarder immediately began charging a fee of US $150 per shipment to the new broker, ostensibly to cover the cost of supplying the new broker with the correct documents and pre-alerts. A KPI brought this cost increase to light, and an investigation showed that the same amount of work and information flow occurred in the forwarder's workload whether the broker was owned by the forwarding company or not. Without the appropriate KPI to highlight this change in cost for a single lane, the extra charge could have become an embedded, hidden, and potentially permanent cost.
2. Goods were being moved from the USA to Dubai via Europe, where they were transshipped from one vessel onto another. The KPIs in place, particularly for cycle time, showed varying delivery times, and hence, problems for the customers. An analysis of the cycle time at the service-provider level showed that the leg from the United States to Europe was very reliable, with a small standard deviation, while the leg from Europe to Dubai was very unreliable and showed a large standard deviation. This was especially concerning, as the same ocean carrier performed both legs of the move, so there should not have been such variations.
Further process analysis turned up two influencing factors. The first was that the leg from the United States to Europe was a high-volume trade lane for the carrier, with all of its European and Middle East freight moving through it. As a result, the U.S.-based lane manager paid a lot of attention to this lane's performance (in terms of cycle time and its variation, measured as the standard deviation) and the level of service was high.
The second was that the leg from Europe was a low-volume lane for the company that only handled freight destined for Dubai. Accordingly, its lane manager in Europe treated its as a low priority. Once the shipping line understood that its customer was measuring its performance on the entire journey from the USA to Dubai, the company ensured that the two lane managers worked together, and service improved dramatically.
Incentives and penalties
In order to drive continuing improvement across the logistics chain, there must be some financial incentive for achieving improvement against the KPIs as well as some financial penalty for deterioration of the economic value delivered. Without both of these, there will be no reason for companies to innovate and improve, or even to simply maintain the standards.
Incentives must be based on improved economic value (including cost and working capital) so the logistics chain will continue to increase the overall value of its contribution toward achieving business objectives. When properly designed, the incentives will be large enough that the partner companies will work at them without continual pressure from the company that is controlling the move.
It's important to recognize that the aim of penalties (really another type of incentive) is to encourage the parties involved to quickly fix a problem and improve, not to punish anyone. Make the penalty too high, and the affected company may weigh the value of the contract and choose to terminate, leaving the whole logistics chain in disarray. For these principles to succeed, the entities involved in the incentive and penalty program must be considered as long-term partners, and mutually working to improve the KPIs must provide value to each party.
The following is merely one example of a way to manage penalties and incentives; companies can develop their own depending on their business and industry.
Each quarter, the KPI improvement by each entity could be measured, and the largest portion of the value of that improvement (say, 50 percent) would accrue to the party that is paying for the move. The remaining 50 percent would be split among the service providers in proportion to the improvement each one has made. Similarly, any entity in the logistics chain whose performance declines could pay for a team of people from the shipping party plus representatives from the immediate upstream and downstream entities to visit the organization with the problem and help it to significantly improve. Of course, penalties must be applied if the performance of the company in question continues to deteriorate by more than, say, 10 percent, or it exhibits deterioration for more than two quarters. These penalties can be calculated from the economic value generated from the KPIs.
These measurements and the associated incentives or penalties must be included in the relevant contracts, so that all the entities are clear on their responsibility for the total success or failure of the logistics chain. Any party that does not want to abide by or accept these measurements should not be considered a long-term partner.
Simple, standard, and necessary
To be effective, key performance indicators and any associated incentive program do not have to be complex. However, whichever company controls the logistics chain should start by making sure that all members agree on three things: the definition of what a KPI is, how it relates to the entire chain's business objectives, and the criteria for applying practical and effective KPIs.
Once that has been established, a company can implement three simple KPIs for an international supply chain: the lane cost, working capital involved, and penalties for excessively late deliveries. These can be used as three separate indicators that are compared to a baseline, or they can be combined to create a single index with a breakdown for each of the different entities that participate in the logistics chain. The measurements are practical and simple to put into effect, and they result in an economic value for each lane. This economic cost also allows a measure of payment for improvements or penalties, which can be very valuable. The ultimate goal is to make sure that value is always added and that improvements are continuous.
Notes:
1. W. Pienaar and J. Vogt, Business Logistics Management: A Supply Chain Perspective (Oxford: Oxford University Press, 2012).
2. Terrance L. Pohlen, "Supply Chain Metrics: Linking Performance With Shareholder Value," CSCMP Explores..., Vol. 2, 2005.
