Amazon recently published an article on a new approach to optimizing delivery routes. You may have already heard about the traveling salesperson problem, which is what computer scientists call NP-hard—meaning that it is very hard.
What I find fascinating about Amazon’s article is that it lays out a very straightforward approach to a complicated problem: start small, then expand. Amazon is a massive company with almost 2,000 data workers, but its approach works across scales. Companies without the resources of Amazon can use the same methodology when testing and experimenting with smaller regions of their network to prove efficacy before expanding.
By focusing on this efficient approach, many companies can achieve massive gains using a combination of machine learning algorithms and heuristics based on business intuition. For example, a software company without any cross-sell initiatives can benefit massively by using simple algorithms—prioritizing speed to action over perfection. Once a production system is in process with a process to synchronize sales, marketing, and data, better algorithms can be used to improve incremental gains.
I truly enjoy prototyping these innovative approaches during diligence to support company valuation. Imagine quickly building a duct-taped version of a model, estimating lift, and incorporating them into a company’s valuation. After the deal is closed, the model can be iterated and undergo rapid iterations with the rest of the company to actualize the gains. No resources are wasted on theoretical work that may never be utilized.
Reach out if this is of interest to you. I’m happy to chat!
