“If you don’t measure it, you can’t improve it”. Sure, there are exceptions. But generally speaking, to improve something, it needs to be measured. Besides, how does one know how well the improvement works unless there is a baseline and a current measure to compare?
But just as important, and often overlooked, is what the KPI represents. Not only does a properly designed KPI represent what is being measured, it must measure something that moves then entire organization forward. If the KPI is to reduce cost by reducing touches or labor content, you must ask yourself if payroll is going to be reduced. Freeing up productive time is only a cost savings if the labor leaves the building or is used elsewhere generating revenue.
For instance, measuring pieces per pick will tell an organization how many pieces a picker gathers each trip into the warehouse or into the forward pick area. While this is often used as a measure of picker performance, it really tells us more about order density and scheduling than picker performance. Yet, a former client used that measurement to determine the performance of pickers.
The problem is that pickers would dig through tickets to find the biggest one. That is a waste of picker time and ruins any work being done to prioritize work. In the end, that KPI did nothing to incentivize picker “efficiency”. Picking the best tickets would allow pickers to take unscheduled breaks and generally drag around for long periods.
More damaging, though, was the fact that this measurement incentivized the wrong activity and ignored a more important aspect of the picking operation. The bigger problem was that pack stations would run dry for long periods awaiting picks to arrive. In our way of looking at things, running a pack station dry for even 30 seconds is a big deal.
This client also had a process set up where the picker would pick somewhere between 5 and 30 orders per pick ticket. Picking 30 single piece orders would take as long as picking 15 orders with 4 pieces each. Each pick ticket would have items spread across the entire warehouse, even if no individual order needed items from more than one rack.
Once the pick ticket was picked, pickers would sort the items from their pick cart into a number of bins matching the number of orders on the ticket. This was because their ERP/WMS didn’t allow for picking multiple orders on a single pick ticket unless all items were picked together, then separated.
Our solution was simple. Assign a bin to the order at release. Print that bin number on the pick ticket and the order ticket. We then limited to 6 the number of orders, not lines or pieces, on each ticket. As the picker “bulk picked” the ticket, they would drop the items into the bins numbered 1 to 6 on a single cart.
We also created a tool to release orders together that picked from the same rows in the warehouse. If the first order on a ticket came from Row 1, then the rest of the orders would come from Row 1 or close by.
We changed the KPI to be pieces per hour and orders per hour and randomly assigned tickets to pickers. We eliminated both digging through tickets for larger picks and the sort at the end of a pick run.
By creating a higher turn, lower volume pick cycle, we ensured pack stations a more steady flow of orders, thus minimizing the stalling due to a dry buffer. We also ensured that pickers were not stacked up sorting through already picked product and choosing pick tickets, thereby eliminating stalling in the picking aisles.
The result was a 300% increase in picking and packing capacity with the same headcount. While we prefer to think of this move as an increase in capacity and better order fulfillment (customer delightment), it could also be thought of as a reduction in operating expense since only 25% of the previous headcount is required to pick the same pre-change volume.
But from a KPI standpoint, it shows the importance of picking the right thing to measure. Had we stuck with the original KPI, the change would have showed a massive decrease in pieces per turn. It would have been seen as a negative move. Imagine a four-fold increase in fulfillment capacity at the same cost being seen as a negative impact!
While many KPIs are not as obviously flawed as this example, there are probably some in any organization that are incentivizing the wrong activity. One I look at often is inventory turns. While getting inventory turns up is generally a great idea, doing so by risking sales is often the result. The objective should always be to reduce inventory not needed to fulfill orders. Without digging into sales volatility and order patterns by-item/by-customer, pushing turns up will inevitably lead to stock-outs. Right sizing inventory based on order patterns considering volatility is the correct approach. More on how to do that in another post.
Key Performance Indicators are fantastic if chosen well, designed correctly and understood by those being measured. If they incentivize improved performance for the business, as shown in increased throughput or reduced operating expense, they are valuable. Otherwise, you may be using resources to make your business less efficient.