Algorithms: Good Enough for Business Purposes

The first step to solving a problem is often deciding whether it is worth solving at all! We are often involved with the first proper e-commerce store for a lot of small merchants and when we start talking about shipping, pricing and discounting rules, we’d clearly need to write a sophisticated reasoning engine just to start to approximate the way they do business. We have one end client that actually builds and weighs the order before pricing shipping using common sense to select the right combination of boxes and shipping carrier to get the best price. And this is for $100-$300 retail orders where the shipping usually runs $5 to $25!

Of course, they could just add the shipping price at the end, but most people ordering small items online for themselves (it’s a B2C company) want a fixed cost before they hand over their credit card details.

While it sometimes makes sense to write complex, custom shipping, pricing and discounting rules, often the cost of specifying, coding and implementing such rules would exceed the likely first couple of years net profits for smaller merchants. Because of that we typically recommend some standard types of algorithms we’ve developed in the past that can be parameterized to meet most needs and that will allow them to get a site live and working in a fraction of the time for a fraction of the price.

We pride ourselves on the ability to generate pretty much whatever the end customer wants, but sometimes it is worth thinking whether what they are asking for is really in their best business interest. It amazes me how many small merchants are hung up on losing $1 here or there on shipping for a particular item rather than just focusing on making their shipping and pricing rules work out in the aggregate.