Metric trees for logistics and supply chain

Logistics measurement is harder than measurement in most other functions, and the reasons are structural rather than technical. Three characteristics make logistics metrics particularly challenging to manage.

First, logistics spans organisational boundaries. A single order touches procurement, warehousing, transportation, and often third-party carriers or fulfilment partners. Each function has its own systems, its own incentives, and its own definition of "on time". The warehouse measures on-time despatch from its dock. The carrier measures on-time delivery to the destination. The customer measures on-time arrival at their door. These are three different metrics with three different owners, and a failure in any one of them produces the same outcome: a disappointed customer.

Second, logistics metrics are sequential and multiplicative. If the warehouse picks the right product but the carrier delivers it late, the order fails. If the carrier is on time but the warehouse picked the wrong SKU, the order still fails. The overall success rate is the product of the success rates at each stage, which means that even modest failure rates at individual steps compound into significant end-to-end failure. A warehouse with 98% pick accuracy and a carrier with 95% on-time delivery produces an end-to-end success rate of just 93.1%, well below what most customers would consider acceptable.

Third, logistics operates under extreme variability. Demand fluctuates seasonally, daily, and even hourly. Weather disrupts transport schedules. Supplier lead times shift. Labour availability changes. Unlike a software team that can control its inputs and environment, logistics teams must deliver consistent outcomes in the face of constant external disruption. This variability means that static targets are often misleading. A warehouse that achieves 99% order accuracy in January and 94% in December is not getting worse at its job; it is being overwhelmed by peak-season volume. The metric tree needs to account for this by separating capability from load.

On-Time In-Full (OTIF) is the most widely used measure of logistics performance, and for good reason. It captures the two things customers care about most: did the order arrive when promised, and did it contain everything that was ordered? OTIF is calculated as the percentage of orders that meet both conditions simultaneously. An order that arrives on time but is missing items does not count. An order that is complete but arrives late does not count. Both conditions must be satisfied.

This binary nature makes OTIF a demanding metric. A business with 96% on-time delivery and 97% in-full delivery does not have 96.5% OTIF. It has, at best, 93.1% OTIF (assuming the failures are independent), and potentially lower if the same orders tend to fail on both dimensions. This multiplicative relationship is precisely why OTIF needs a metric tree: you cannot improve the composite without understanding which component is dragging it down, and you cannot improve a component without understanding which operational driver is causing the failure.

The tree reveals that on-time delivery is driven by three sequential stages, each with its own failure modes. Order processing time covers the gap between order receipt and the warehouse receiving the instruction to pick. Delays here often stem from manual order entry errors that require correction, or credit holds that pause fulfilment. Warehouse despatch timeliness measures how quickly the warehouse converts a pick instruction into a shipped parcel. Transit time reliability captures carrier performance from pickup to delivery.

In-full delivery decomposes into two distinct problems: was the product available to ship (inventory availability), and did the warehouse pick the correct items in the correct quantities (pick accuracy)? These have fundamentally different solutions. Inventory availability is a planning problem solved by better demand forecasting and safety stock policies. Pick accuracy is an execution problem solved by better warehouse processes, slotting strategies, and scanning technology.

The practical value of this decomposition becomes clear when OTIF drops. Instead of launching a general investigation, the tree tells you exactly where to look. If on-time delivery fell but in-full delivery held steady, the problem is speed, not accuracy. If transit time reliability dropped but warehouse despatch timeliness remained constant, the problem is with carriers, not warehouse operations. Each branch points to a different owner, a different root cause, and a different intervention.

The warehouse is where logistics promises are kept or broken. It is the conversion engine that turns inventory into fulfilled orders, and its performance directly determines both OTIF and cost to serve. Yet many warehouse operations track dozens of metrics without a clear hierarchy showing which ones drive the outcomes that matter. A metric tree for warehouse operations organises these metrics into a coherent structure that connects floor-level activity to business-level results.

The two outcomes that matter most for a warehouse are throughput (orders processed per hour or per shift) and accuracy (percentage of orders shipped correctly). These are not independent: pushing throughput too hard degrades accuracy, and pursuing perfect accuracy slows throughput. The metric tree makes this tension visible and manageable.

A common mistake in warehouse measurement is treating labour productivity as the primary metric. Lines picked per hour or orders shipped per full-time equivalent are important, but they are efficiency metrics that can be gamed by cutting corners on accuracy or by cherry-picking easy orders. In a metric tree, labour productivity sits below throughput and is balanced against accuracy metrics on a sibling branch. This structure prevents the classic warehouse failure mode: a productivity drive that improves lines per hour by 15% while increasing error rates by 3%, resulting in a net negative outcome when the cost of returns, re-shipments, and customer complaints is factored in.

Inventory accuracy deserves special attention in the warehouse metric tree. The gap between system-recorded inventory and actual physical inventory creates downstream failures that appear as stockouts, mispicks, and late shipments. Cycle count accuracy, the percentage of SKUs where the physical count matches the system record, is a leading indicator that predicts many of these downstream problems. Warehouses with cycle count accuracy below 97% typically experience two to three times the order error rate of those above 99%.

Transportation is often the largest single cost in the logistics budget, typically representing 50 to 70 percent of total logistics spend. It is also the stage most exposed to external variability: weather, traffic, carrier capacity, fuel prices, and regulatory changes all affect performance. A metric tree for transportation needs to balance cost efficiency against service reliability, and it needs to account for the fact that much of the execution is performed by third parties whose incentives may not perfectly align with yours.

The metric tree for transportation connects these metrics in a way that reveals root causes. Freight cost per unit shipped decomposes into three primary drivers: the mix of transportation modes (ground, air, ocean, rail), the rates negotiated with carriers in each mode, and the load utilisation achieved. A spike in freight cost per unit could be caused by a shift toward more expensive modes (perhaps driven by late orders requiring expedited shipping), by carrier rate increases, or by falling load utilisation (smaller, more frequent shipments).

Carrier on-time performance is the transportation metric that most directly affects customer experience. It decomposes into carrier pickup reliability (did the carrier collect the shipment when scheduled?) and in-transit performance (did the shipment move through the network on time?). When carrier performance drops, the metric tree helps you distinguish between carriers that are consistently underperforming (a procurement problem requiring carrier replacement or contract renegotiation) and specific lanes or seasons where all carriers struggle (a network design problem requiring alternative routing or mode shifts).

One of the most valuable insights from a transportation metric tree is the connection between expedited shipping and upstream failures. When the proportion of expedited shipments rises, it almost always signals a problem earlier in the chain: late production, inventory shortages, or delayed order processing. Expedited shipping is the most expensive way to compensate for these failures, and the metric tree makes the cost of that compensation visible by connecting mode mix to the upstream drivers that influence it.

Inventory is where supply chain planning meets physical reality. Too much inventory ties up working capital, consumes warehouse space, and creates obsolescence risk. Too little inventory leads to stockouts, lost sales, expedited shipping, and disappointed customers. The challenge is not simply to hold "the right amount" of inventory but to hold the right amount of the right products in the right locations at the right time. This multi-dimensional optimisation problem is why inventory metrics need a tree structure rather than a single KPI.

The traditional measure of inventory efficiency is inventory turns: the number of times stock is sold and replaced over a period. Higher turns generally indicate more efficient use of capital. But turns alone can be misleading. A business could improve turns by eliminating safety stock, which would simultaneously improve the financial metric while degrading service levels through increased stockouts. The metric tree prevents this by placing inventory turns alongside service-level metrics, making the tradeoff explicit.

1. 1

   Inventory turns and days of supply

   Inventory turns measure how frequently stock cycles through the business. Days of supply translates this into a more intuitive measure: how many days of demand could the current stock satisfy? Both metrics should be segmented by product category, because aggregated turns hide the reality that fast-moving products may turn 20 times a year while slow-moving tail items sit for months.

2. 2

   Fill rate and stockout frequency

   Fill rate measures the percentage of customer demand that can be satisfied immediately from available stock. Stockout frequency counts how often a SKU reaches zero availability. These are the service-level counterparts to inventory turns: they measure the consequences of lean inventory. In the metric tree, they sit on a sibling branch to turns, making the efficiency-versus-service tradeoff visible.

3. 3

   Demand forecast accuracy

   Forecast accuracy, measured as the mean absolute percentage error (MAPE) between forecast and actual demand, is the leading indicator that drives both inventory efficiency and service levels. Poor forecasts create excess stock of products that do not sell and shortages of products that do. Improving forecast accuracy is often the single highest-impact lever for improving both turns and fill rate simultaneously.

4. 4

   Inventory accuracy

   The gap between what the system says is in stock and what is physically on the shelf. Inaccurate inventory records cause phantom stockouts (the system shows zero, but stock exists) and phantom availability (the system shows stock, but the shelf is empty). Both destroy fulfilment performance. Cycle count accuracy above 99% is the threshold where inventory record errors stop being a significant source of order failures.

5. 5

   Obsolescence and write-off rate

   The percentage of inventory value written off due to expiry, damage, or obsolescence. This is the cost of overstocking. In the metric tree, it sits alongside carrying cost as a consequence of holding excess inventory, balancing the pressure to increase stock levels for better fill rates.

The last mile is the most expensive, most visible, and most emotionally charged segment of the logistics chain. It typically accounts for over 50% of total delivery cost while covering the shortest distance. It is also the only part of the supply chain that the end customer directly experiences. A parcel that moved flawlessly through warehouses and linehaul networks for three days but is left in the rain on the doorstep is remembered as a delivery failure, not a 99% success.

Last-mile metrics need to capture three dimensions: cost efficiency, delivery reliability, and customer experience. These dimensions interact in ways that a flat dashboard cannot show. Reducing delivery cost by widening delivery windows frustrates customers. Tightening delivery windows increases cost and reduces first-attempt success rates. Offering free re-delivery improves customer experience but inflates cost per successful delivery. A metric tree organises these tradeoffs so that improvement in one dimension is always evaluated against its impact on the others.

The first-attempt delivery rate deserves particular attention because it is a leverage point where small improvements compound into significant financial and customer outcomes. Industry benchmarks suggest that a failed first delivery attempt costs three to four times more than a successful one when you factor in the re-delivery attempt, customer service contacts, and the risk of the customer cancelling or returning the order. A logistics operation delivering 1,000 parcels per day with an 85% first-attempt success rate is absorbing roughly 150 failed deliveries daily. Improving that rate to 92% eliminates 70 of those failures, directly reducing cost and improving customer experience.

The metric tree for first-attempt delivery rate reveals where to intervene. Address accuracy failures require better address validation at the point of order. Recipient availability failures can be addressed through more precise delivery time windows, real-time tracking notifications, or safe-place delivery options. Delivery window failures point to route planning problems, driver capacity issues, or unrealistic promise times set by the commercial team. Each cause has a different owner and a different solution, and the tree ensures that improvement efforts target the actual cause rather than the most visible symptom.

Logistics teams often struggle to articulate their impact in terms that resonate with the wider business. Warehouse throughput, carrier on-time rates, and pick accuracy are meaningful to operations professionals but can feel abstract to a CFO or a chief commercial officer. A metric tree that extends from logistics drivers upward to financial and customer outcomes solves this translation problem and positions logistics as a strategic function rather than a cost centre.

The connection works through two primary paths: the cost path and the revenue path. On the cost side, logistics metrics feed directly into cost of goods sold and operating expenses. Freight cost per unit, warehouse cost per order, inventory carrying cost, and returns processing cost all appear on the income statement. Improving these metrics drops straight to the bottom line. On the revenue side, logistics performance affects customer satisfaction, repeat purchase rates, and brand reputation. A business that consistently delivers on time and in full retains more customers, generates more positive reviews, and can command premium pricing.

The most effective logistics metric trees include a financial layer at the top that makes these connections explicit. Cost to serve, the total cost of fulfilling an order from receipt to delivery, is a powerful bridge metric because it aggregates all logistics costs into a single figure that can be compared to the revenue and margin generated by each order, customer segment, or channel. When the logistics VP can show that reducing cost to serve by 8% through warehouse automation and carrier optimisation will improve gross margin by 1.5 percentage points, the conversation with the executive team shifts from "how do we cut logistics costs" to "how do we invest in logistics capability".

Equally important is the connection between logistics metrics and customer lifetime value. A customer who experiences a late delivery is significantly more likely to reduce their purchase frequency or switch to a competitor. The metric tree can trace this from the operational failure (a carrier missing a delivery window) through the customer experience (a late delivery notification) to the financial consequence (reduced repeat purchases and lower lifetime value). This end-to-end visibility is what transforms logistics from a back-office function into a competitive advantage.