We recently completed our warehouse renovation just in time for Memorial Day. I snapped a quick photo of the new space, which you can see we’re already utilizing:
The big advantage of this new space is that we have almost 6,000 sq-ft more to dedicate to products. That means we can space products out better and reduce the usage of separate overstock areas. It also keeps the aisles and shelves more organized.
Deciding on the new shelving layout and product flow was pretty easy. We consulted with our employees, came up with an initial design, did another small set of tweaks, and then executed our plan as soon as we had the space available to us.
The resulting layout ended up adding quite a bit of complexity to our picking operation. We’ve long used the inventory zone system I wrote about in 2010 (and slightly improved upon in 2017). It became clear pretty quickly that this would not work for the new layout. Let me illustrate.
Before, we had a relatively simple layout. The shelves faced each other to allow for quicker picking, easier stocking, and the ability to have a team doing each at the same time without interfering with each other.
When picking, employees would simply grab a box and the packing slip would route them through the warehouse in zone order. So from Aisle 1 to Aisle 6, and within each Aisle from Unit 1-18. Pretty simple, and it was pretty effective for us.
However, with the new layout below, you can see pretty quickly that problems arise with that simple approach. With the introduction of a cut-through aisle, employees can now enter an Aisle from the Front, Middle, or Back.
If the Units outlined in red above had items in an order, the ideal route would look something like this:
Which isn’t in Aisle/Row + Unit order anymore necessarily. It’s much more complex. Almost like a GPS system.
We couldn’t think of any shortcuts, so I set out to build an entire routing optimization system to minimize steps. Essentially, at each step it’s looking at every other item on the packing slip, calculating the number of steps to get to that item by Front, Middle, and Back aisle entry, sorting all of those step counts, and then picking the item and entry point with the least amount of steps. Then for the next item, it does it all over again. It has a few other rules to minimize steps and ensure good flow, like it won’t bounce between Aisles (Aisle 1 needs to be complete before Aisle 2, Aisle 2 before Aisle 3, etc).
Unlike most projects, the volume of possible scenarios made it tough to test. I built the system in December and January, unsure of when we’d need to use it. By May we’d made some tweaks, and I’d run as many test simulations in as many ways that I could think of, but I really had no idea how it would work in practice, and no way to test it out slowly. The launch would be all or nothing during our peak season. The thought of this going wrong was very stressful.
Thankfully it all went well. There was only one minor bug that took a few minutes to fix on Day 1, and other than that it’s gone pretty smooth so far.
A system like this is really key for us. We strive for fast and affordable shipping on every single order. The “fast” part of that requires that we have very efficient warehouse operations. Our employees don’t waste steps, which is good for them and good for us. We’re able to ship out a lot of orders in a very short time, with minimal mistakes, and a pretty small crew.