Unintended side effects

Starbucks is in big trouble. In recent years – especially in the US – placing orders online for subsequent pick-up in its outlets has grown substantially. This has resulted in long queues that also irritate customers who are ordering on the premises. And on top of that, the online customers order products with lower margins than customers who order in store.

Last week I saw results of a study for a large European supermarket chain where the number of in-store customers has dropped sharply now that store employees are constantly walking up and down the aisles picking (loss-making) orders for online customers. The (profitable) in-store customers no longer like doing their shopping there.

In a very different example, a manufacturing company we work with has largely automated the order entry and exit based on the current delivery status. That has led to so many messages to suppliers that many of them have decided to simply ignore them all. And recent research in our lab has shown that human demand forecasters consciously adjust their forecasts when they know that the inventory manager is an algorithm rather than a person. I think many vendors of order management software either don’t know that or don’t tell people.

Ultra-slow tsunami

Why do companies still keep underestimating these unintended side effects? The good news for project managers and consultants is that many unintended side effects are not apparent immediately after implementation. In supply chain projects, they are a kind of ultra-slow tsunami that only becomes visible after many months – or doesn’t become visible at all, because there is no thorough evaluation of changes based on advanced econometric techniques. Those project managers and consultants have often left before it becomes apparent. And during the project, it is often a case of ‘overconfidence bias’ and ‘confirmation bias’: managers are too confident in what they intend to achieve with a project, and have not surrounded themselves with sufficient naysayers.

For digitalization projects that companies are now deploying, however, much evidence of these unexpected side effects has been provided in the scientific literature in recent years. A lot of studies are now being done on the application of artificial intelligence, gradually revealing what the unintended side effects might be. These are not popular with technology advocates, but very important to take note of in order to achieve implementations that are also successful in the long term. If there are no naysayers in your company, they can be found online… scientifically validated.

Jan Fransoo, Professor of Operations and Logistics Management at the Tilburg School of Economics and Management