Divide and Conquer

The Pizza Problem

I love pizza and frequently order from a shop that guarantees to deliver a pizza to my doorstep within 30 minutes from placing the order. I’ve always wondered what it takes for them to receive an order, prepare a pizza, get it packed with sides, travel to the destination locality, search for the correct house and then ring the bell – all within 30 minutes – every single time.

Usually it takes around 12-14 minutes to prepare a pizza and they cannot cut short that time. Packing the food and handling the billing will add another 2-3 minutes. Then the delivery person has to travel a certain distance which he or she cannot fly and cover…yet. (I live in Bangalore and the traffic here is maddening). So I wonder just what are they doing differently to other pizza shops to make these timely deliveries?

One fine day I happened to visit their store and peeped into their back room. There was a framed map of my area and on the glass they had drawn straight lines in a grid like this:


Curious about the map and those partitions, I asked the store manager about the specifics of the map. He told me that he has divided his delivery area into imaginary zones and assigned each to his delivery boys. The delivery boys are supposed to know every main road, every cross road and every landmark in their zone. That way, when they are out for delivery in their own assigned zone they do not lose time looking for their destination. The manager said that it is impossible for six delivery boys to know full six square kilometers of the shop’s delivery area, but it is very much possible for one single person to know one square kilometer within that area. Now, all he needs to do is ensure that the right delivery person is assigned to the order placed from the individual zones.

It occurred to me that their problem is similar (if not the same) to the one in a typical warehouse, i.e. to fulfill an order quickly, and efficiently. And for a similar problem why can’t we have a similar solution? Why can’t we too ‘Divide and Rule’?

Warehouse Partitioning

What if we partition the warehouse storage locations into zones and assign those zones to individual pickers? Take the example of a warehouse which is 10,000 square feet, has roughly 4,000 storage locations and eight pickers cover those 4,000 locations. Now it is impossible for eight pickers to know where products are located in 4,000 locations. But it is very much possible for one picker to know what is where in his quota of 500 locations. This does not mean that the picker will have to know exactly which box contains what, but he can have a rough idea in his mind that the items in his zone are in so and so order, the smaller SKUs are kept here and next to them are the fragile ones… and so on.

This will solve two problems. The first being the quickness in identifying an item, which can be a lot quicker than earlier when the picker was dependent on the system to guide him. It is like searching for a house in a neighborhood with the help of GPS (which more often than not will take you to the correct house) vs. searching for a house in your own neighborhood (to which you are well versed of and you do not need a GPS). Obviously you will find the house (read location) faster when it is in your own neighborhood.

Secondly, one of the major pain points for the workforce in a warehouse is that they have to walk – a lot. I read an interview with a picker who worked in a large Amazon warehouse and he said he had to walk over 30 miles a day (within the warehouse off course). This might be a one off case but we all know that a picker typically has to cover large areas. So if we go for ‘warehouse partitioning’ approach, the distance travelled by the pickers can be reduced drastically as well. Which in turn will keep them happy and less tired, which in turn will add to more productivity for the warehouse.

There can be some challenges with this approach. What if a picker who is familiar from his zone is unavailable? Then the business will run as it does today, depending on the systems available. Or what if the picking traffic for one zone is more than the other? A main reason for sensible partitioning and not just a square grid like that in the pizza store. There can be myriad other similar problems, but I believe that with due diligence those problems can certainly be tackled.

There can be an argument against this approach saying that we are going back to old / manual ways. Well first of all, not everything is bad with “old ways.” And secondly, the suggested approach will be a hybrid between the old and the new ways of doing the job!

Would you like to identify new opportunities to optimize your warehouse operations? Contact JDA today to learn how we can help you boost productivity and service!

  1 Comment   Comment

  1. To the main question that was asked “Why can’t we too ‘Divide and Rule’?” ( in warehousing ) – Grouping the picks of an order by Zone or Zone picking strategy where some operators/equipment work within defined zones as far as I can tell has been around for a very long time and would be applicable to execution by manual labour or automation. In fact most of the practices in store operations such as the one you mention are known to have evolved from Manufacturing and Warehouse management. Zone picking introduces an additional lead time to re-consolidate the order not to mention increased probability of errors over simple Order Picking where an operator would complete all picks of an order by traveling to all required locations in the warehouse. Also if you bring in the objective of maximizing resource utilization within zones then only a certain volume of picks per order can generate work for all the zones so we are now looking at something similar to the classic assembly line balancing type problem. A key objective like you mentioned is to minimize the total lead time to pick an order but we also looking at the trade offs with other optimization objectives and constraints. For example Zone picking strategy could work well for large store orders but may turn out to counter productive from resource utilization perspective for small E commerce orders or for spare parts with few lines per order, small order quantities, wide dispersion of picks in warehouse locations from order to order etc. The challenge for warehouse science practitioners would perhaps be to come up with credible answers to follow on questions that have been asked widely in the warehousing community such as ‘How do we further divide and conquer picking lead times for multiple types of order profiles in an omni-channel distribution environment ?’.


Leave a Comment