Dr. Michael Watson, one of the industry’s foremost experts on supply chain network design and advanced analytics, is a columnist and subject matter expert (SME) for Supply Chain Digest.
Dr. Watson, of Northwestern University, was the lead author of the just released book Supply Chain Network Design, co-authored with Sara Lewis, Peter Cacioppi, and Jay Jayaraman, all of IBM. (See Supply Chain Network Design – the Book.)
Prior to his current role at Northwestern, Watson was a key manager in IBM's network optimization group. In addition to his roles at IBM and now at Northwestern, Watson is director of The Optimization and Analytics Group.
By Dr. Michael Watson
December 5, 2012
Same Day Delivery and Network Design
For Quickly Evolving Markets, Building Simple Models and Testing a lot of Different Ideas Might Be the Best Option
As Dan Gilmore pointed out in a recent First Thoughts column, e-commerce and multi-channel commerce is booming.
Dr. Watson Says:
We don't know how same-day will evolve and how these questions will get answered. But, network modeling can help you understand the parameters.
And, as part of this boom, a recent Wall Street Journal article highlighted the many different retailers trying same-day delivery. This article highlighted some business models that won’t scale—sending a courier an order, having that courier go to the store and buy the item and then deliver to the customer.
Walmart and Amazon have probably received the most attention for their same-day delivery efforts. Amazon is taking advantage of their e-commerce brand, their distribution network, and their lockers. Walmart is taking advantage of their stores and some unique customer insights (people want to order on-line, but pick up and pay in cash at the store).
It is impossible to tell how this will evolve:
Retailers need to define what “same day” means. For example, does it mean within 1, 2, or 4 hours? What are the cut-off times? And, does the retailer offer different service depending on your address?
Retailers need to figure out if they can charge for same-day delivery and if they can replicate successful same-day models (like pizza).
And, I’ve heard that UPS, FedEx and USPS are trying to get into this market. These carriers are already going to the home. Can they coordinate with the retailers to make same-day deliveries? It is similar to the old concept of merge-in-transit except done at a micro-level: you can imagine scenarios where the retailer actually delivers to a single UPS truck that is already on its route. Of course, this requires extreme coordination.
We don’t know how same-day will evolve and how these questions will get answered. But, network modeling can help you understand the parameters.
Most of us think about network design helping locate warehouses across the country.
But, exactly the same concepts apply to a local market. You want to think about your customer and demand points as census tracks, city blocks, or an artificial grid of square miles (or square half-miles). Your distribution points would be points where you keep inventory like your stores.
The analysis would help you determine how much coverage you could offer or what your realistic cut-off times should be.
Final Thoughts
When doing a network model like this for a market that is evolving quickly, we have found that it is best to build simple models and test a lot of different ideas. You looking for the model to tell you what has to happen to make this business model work for you. For example, how many locations do you need? How many potential customers can you cover? How do changes in the time to deliver impact the solution? You still have a lot of work to implement a solution. The network design model helps you narrow the list of potential solutions.
Recent Feedback
What makes something of this scope more ripe for simple and different than any design?
I like that you address some of the granular detail with the "what does 'same day' mean" questions. This so often seems ignored in the media coverage of same day delivery. The micro level merge in transit paints an interesting picture; especially from an operational accountability perspective.
Only touched on in the last paragraph, I don't see an argument supporting simple models and different ideas.
Most network designs start with the Holmesian approach of eliminating the impossible to leave only the possible from which to craft a solution. This may not always be done in a formal model but it is done at some level. The quick churn through of several possibilities sets the stage for the real work. The only difference here seems to be scale, essentially from thousands of miles to a few.
