So, following the two latest blog posts covering e-Commerce topics, it is time to take a look at fulfillment metrics. The focus will be on the warehouse side of things, simply because the warehouse is at the center of every e-Commerce, or any other logistical, operation. To better understand the importance of the KPIs, or metrics, and there are plenty, let’s have a first look at why the warehouse is so important.
The Importance of Warehousing
If we look at any product flow, whether it is for production, retail or e-Commerce, it becomes clear that every single unit is going through one, and in most cases multiple, warehouses. Warehouses hold the majority of inventory and thus bound capital. Warehouses are also among the most expensive pieces of a supply chain, after ships and planes.
This means warehouses are representing a major cost factor. And every mistake made in a warehouse propagates down the supply chain, directly impacting customers to various degrees. By holding a lot of capital, warehouses are also a crucial element of accounting and balance sheets. Especially since a lot of warehouses are operated by third parties. If you ever went to through a SOX audit, the importance of stock taking and inventory records becomes obvious.
Warehouses are not just holding inventory, they are also directly driving inventory levels. Simply put, the more warehouses there are in a given supply chain, the higher inventory levels will be.
I strongly believe in starting with the customer and work backwards. So how do warehouses impact customers in ways that cannot be recovered by other functions? Or in other words, what do customers expect, and how are warehouse operations affecting these expectations?
Customers expect on-time delivery of the right products at the right time at the right place. Warehouses directly impact the first two of those. Customers also expect their orders to be delivered at the right cost. Warehouses are also directly impacting fulfillment costs. Customers, especially B2B customers, also expect the right information, e.g. serial numbers. Warehouses are again key in tracking that. Same is true for the right quality, as every time a product is moved, damages can occur.
So which KPIs can be used to measure an manage warehouse operations? And even more important, how?
Measuring Defects vs. Measuring Non-Defects
Usually, KPIs are measuring things that meet expectations. On-Time-Delivery is the percentage of shipments being shipped on time divided by all shipments, for example. Let’s take a high performing operations, delivering 99% of orders on time. Everybody would be happy, and the operation is running as planned, right?
Now let’s take the same operation, but measuring the defects. And not in percent, but in DPMO, in Defects-per-million. Now, the same KPI is not called On-Time-Delivery anymore, it is called missed shipments. And instead of everybody being happy about 99%, we are looking at a number of 10,000 for Missed Shipments. Things aren’t looking that good anymore, are they?
This change of perspective will be applied to every KPI in the post, this change of perspective makes a huge difference for operations. And now, let’s have a look at the KPIs.
Customer facing KPIs – The Obvious Ones
Everyone is talking about on-time-delivery. And it is a good one, for sure. For me, it is lacking depth. A good KPI system is set-up in a way to measure interfaces between processes an entities in a clearly defined, standardized way. On-Time-Delivery is not doing that in itself, especially not when only measuring it at the very end of the chain. How are you going to identify the root cause for late deliveries? How are going to come up with solutions? To make it short, you can’t.
From a warehousing perspective, OTD has to be broken down. It has to be possible to clearly measure late shipments caused by the warehouse and those by carriers. Which means the warehouse needs one KPI for missed shipments caused by the warehouse. This KPI takes expected Ready-to-Ship dates and times, based on warehouse cycle times and carrier pick-ups, and measures all shipments missing that Ready-to-Ship event. The other KPI needed now is the more traditional OTD, measuring how many shipments left the warehouse on time. Now, it is clear with a single look at these two KPIs where a delay happened. An what would be a good a benchmark, measured in DPMO for these two KPIs? For the first one, Read-to-Ship, a good value is, in my opinion, around 3,000 DPMO. For the latter, shipments leaving the warehouse late, it should be below 5,000 DPMO.
What other, directly customer facing KPIs are there? Going back to what customers expect, and what kind of warehouse mistakes cannot be recovered elsewhere, that leaves the following:
- Wrong product
- Damaged product
Arguably, there are two ways to measure them. At the warehouse, by measuring how many products are not shipped due to these two reasons, and by measuring customer complaints due to these two reasons. Both ways have pros and cons. And both ways are incredibly fuzzy.
Measuring damages at the warehouse, or rather measuring abandoned picks due to damages, is possible. With the important limitation that only damaged packaging can in any meaningful way be detected at the warehouse. The same is true for wrong products, only the very obvious ones can reasonably be detected at the warehouse. Due to these limitations, I would not measure that at the warehouse. Which means customer feedback is needed.
Customer feedback for damages and wrong products can by itself be hard to tracked systematically. Still, it is the best source to get any data on damages and wrong products. Wrong products pose less of a problem, but damages, when measured at the customer, can have occurred either at the warehouse or during transit. So again, this KPI is a fuzzy one. And then there is the question of which period should be used to record these defects? Recording them during the period the customer complained doesn’t really help, as the defect happened earlier. And in order to measure warehouse performance, the period during which mistakes happen is paramount. What does this mean for this KPI? That it is fuzzy, defect sources are hard to define and the KPI will change over time, because customers complain in one week for a shipment from two weeks ago. Does this mean it doesn’t have to used? Not at all, it just means that it cannot be reliably used on its own and as frequently as missed shipments can. And for any damages that happened during transport? Well, there are usually as many damages recorded under transport that happened at the warehouse than there are the other way round. At least for e-Commerce.
The Less Obvious Ones
Now it becomes interesting, because not all of these KPIs are directly affecting customer experience. And because there will be a lot, if any e-Commerce warehouse has to be managed at the at high performance levels. So which ones are there?
Looking at warehouse operations, the first time a warehouse gets in contact with a customer order is when it is sent to the warehouse. So first thing to measure, and build a KPI around, is by recording any orders that are refused by the warehouse because the WMS doesn’t find sufficient inventory. Obviously, that only makes sense when there is an interface between the warehouse and the system managing the inventory for customer orders. One example would be a 3PL warehouse fulfilling order for a webshop with integrated inventory management. This KPIs gives a very good indication, of how good different systems interact with each other and how good processes are aligned. A good value, again in DPMO, would be under 500. Everything above 1,000 DPMO is a reason to take an urgent look at systems and interfaces.
The second time a warehouse is in contact with a customer order is during picking. So it makes perfect sense to measure how many picks are cancelled, for various reasons. Again, we measure that in DPMO. This KPI also ties back into warehouse damages or wrong products. Further breaking this down by reason codes doesn’t make a lot of sense so. Recording reason codes, if possible, should be done so, to make root cause analysis a lot easier. What reasons are there? Everything from inventory not at the expected location, last item damaged to wrong item. This KPI ties into damages and wrong item shipped as well as into inventory quality. A good value is anything under 1,500 DPMO, preferably under 1,000. Everything above 2,000 is reason to have an urgent and close look at the warehouse in question, especially if orders are getting rejected due to mismatched inventories at the same time.
The Corner Stone of a Warehouse, Inventory
By now we are already using 5 main KPIs per warehouse:
- Shipments missed by the warehouse
- Shipments leaving the warehouse late
- Customer reclamations due to wring or damaged products
- Customer order rejected by the warehouse due to mismatched inventory records
- Cancelled picks
The first two can even have an additional distinction by order type, something I’d highly recommend when you are aiming to, or already are, fulfilling Amazon Prime orders.
But now, we have to measure something very critical as well. Inventory Quality. Bad inventory quality drives costs, makes e-Commerce fulfillment close to impossible and can cause a lot of financial damage. So how exactly can all of that being measured? And is there a way to use one single metric? The answer to the last question is no, one KPI is not enough.
What exactly do we want to measure? We want to measure how good a given warehouse is managing its inventory, its stock. The last of the above mentioned KPIs are already a god, first indicator. The problem is, so, that these two need a customer order to trigger a defect. Which cannot give us a clear picture of what is going on. So we need something else. Three KPIs, to be exact.
All of these KPIs are based on the inventory control, or stock taking, activities of the warehouse. There are quite a few ways to tackle this, which I will cover in a different post, for now let’s assume the warehouse is counting regularly, every week, a certain number of locations. This counting can result in:
- Inventory that is found at a counted location, the inventory value is adjusted up
- Inventory that is not found at the counted location, the inventory is adjusted down
- The number of defective records and the number of records counted
And this are, basically, our three KPIs directly monitoring inventory quality. We monitor inventory adjustments in inventory value being adjusted, once in gross adjustments and once in net adjustments. The first one sums up the absolute values of adjustments in a period. While this is not reflective of the financial impact of these adjustments, it is a much better indicator of inventory quality, since positive and negative adjustments are not offsetting each other. Net adjustments provide the financial impact in a period, with positive and negative adjustments being offset.
The last information coming from stock taking, number of defective records and number of records counted, can be used to calculate, again in DMPO, the number of wrong inventory records. This rate is calculated by dividing the number of defective records by the total number of inventory records in the warehouse. Directly related to cancelled picks and adjustments, this is the prime indicator for inventory quality issues. Values depend on how many records there are an if the warehouse is storing smaller or larger items. For a warehouse with smaller items, a good value would be around 5,000 DPMO. In the case of larger items, the value should be lower, around 3,000 DPMO. That benchmark will again go up for large article an a low number of inventory records, and should around in the long run somewhere between 3,000 and 5,000 DPMO.
So many KPIs, and so little Time
Now we are already at 8 KPIs and we didn’t even touch inbound operations yet. Can that many KPIs be managed? Yes, even on a weekly basis, and even with covering inbound operations and some other things I will cover in a separate post. It does take practice, so, for both, the warehouse and the team managing and monitoring the warehouse. Unfortunately, modern e-Commerce needs that kind of detail. Not having it increases reaction times when things go wrong, defect are detected too late, customer experience suffers. And costs increase. All things we don’t want for our operations, customers and clients.
You will notice the absence of things like order cycle times, perfect order rates or order fill rates. Cycle times are important, but should be used by the warehouse ops team. Using them at a higher level looking at the business in general isn’t necessary outside of business reviews. And in case of 3PL warehouses, productivity is the 3PLs problem, not the clients. Measuring a perfect order rate poses a problem. It is already measured in various KPIs like cancelled picks, adding a compounded KPI would just blur the picture. Measuring prefect orders instead of defects can also, if the rate is high enough, prompt people to settle for a less than ideal situation. And order fill rates are already included in cancelled picks, every pick order is directly linked to one order line. Aggregating this KPI at a higher level would again blur the picture.
Now, I wrote enough about KPIs. At least for part one. Part two will take a look at inbound operations, while part three will take a look at how this set of KPIs can be used on a monthly, weekly and daily, yes, daily, basis.