“Despite hundreds of millions of dollars invested in supply-chain visibility systems in the 90’s and the 00’s, we’re still far from having systems that can automatically diagnose the state of your supply chain on a given day, and take operational actions to save your butt.”
Last post, I talked a bit about the history of supply chain visibility systems, and about how much of the money invested in creating such systems was flushed down the toilet. So why are some of these supply chain visibility dreams—the dream of having a system that can somehow foretell your supply-chain future and intelligently and automatically take action, given the occurrence of some supply chain event—so far from being realized?
1) The systems don’t solve “black swan” supply chain problems – The supply chain issues that cause executives to really lose sleep are the black swans that they haven’t thought of yet, and could never have imagined if they tried. They aren’t losing sleep about what will happen if one shipment is late, which tends to be the problem that many supply-chain visibility systems are designed to solve. They are worrying about huge events that could cause serious supply-chain disruptions. A nuclear accident and tsunami in Japan. A flood in Thailand. The only problem is that we all have no clue what the next massively disruptive event will be. And if fact, as the geniuses on Wall street learned when their models failed to predict the sub-prime melt down, your systems might blind you to the existence of these black swans and make you overconfident of your ability to avoid them, because by definition your systems model the known world, not the unknown events that we have trouble even envisioning. If you’re most concerned about these colossal supply-chain interruptions, most supply-chain visibility systems aren’t going to be much help.
2) Inability to model the supply chain sufficiently well – In order to create some kind of supply-chain monitoring system that can actually detect problems and take actions, you need to have a model of your supply chain. Let’s leave black swan events aside, for the moment. Even then, our supply chain model needs to be mostly correct, and it needs to then be implemented in a piece of software. We are far from understanding our supply chains in enough depth to be able to do this. E.g., if a truck shipment misses a connection with an ocean container vessel, you might want your system to discover this and rebook the shipment on another ocean vessel. Or even better, you might like the system might be able to expedite the shipment somehow and get it to the port on time for the original vessel sailing. At least, that was the theory. Anybody who’s ever been operationally involved in logistics will roll their eyes about the possibility of achieving these things through some automated system. It’s hard enough to do in the real world, much less automate this through software. At best, you might get an email telling you that there’s a problem, in most cases hours after the fact. But if your freight forwarder is any good, they would know about the issue long before you got that email from a supply chain visibility system.
3) Data – Supply chain visibility is a big B2B integration problem. You need to gather timely and accurate data from multiple other parties in the supply chain, pull that data into one system, and make some sense out of it. The technology required to receive and transform this data exists and is getting better every year. You can do it with your own in-house TMS or ERP system, or with a cloud based supply-chain visibility system, or you can get a 3PL or 4PL to help you. The problem remains, though, that the data still isn’t as timely and as accurate as it needs to be. If you look at typical EDI in-transit status messaging from US-based trucking firms, these are sent hours (and in some cases days) after the events that they describe occurred. People imagine that these messages might have the accuracy of Fedex and UPS package delivery status messages. Sorry. Most trucking firms can’t do that. They may have cab-based systems for tracking trucks, but they aren’t oriented towards providing data to their customers–their purpose is to provide operating data for the carrier. I’ve implemented in-transit visibility systems in Europe and Asia and the data is much worse there. Whereas in North America, where in-transit tracking data is merely shitty and late, in Europe and Asia (apart from giants like DHL), you find yourself in a discussion with the carriers that makes you feel like you are inventing in-transit messaging from first principles. (OK, and now try doing that with a carrier in Japan, when the meetings are held in Japanese, with a translator!) The technology exists to vastly improve on this–satellite tags, RFID. And there are pockets of the supply chain where these technologies have been implemented. But deployment in the “real economy” has been slow.
4) People are the best “tools” for managing exceptions—Reacting to exceptions is exactly when you want humans involved. Do you really want a machine to authorize chartering a plane to get a shipment there on time just because your visibility system discovered that the trucking carrier dropped the ball on it? A good human being is going to be able to think creatively about the problem and take action. A machine? Maybe one day. Machines are great at automating routine tasks. But humans are still the best exception management tools invented. So your best bet is to present the information from your supply-chain visibility system to a human being, and let them handle the exceptions, rather than try to automate this.
OK so now that I’ve depressed you about the inability of supply-chain systems to automatically diagnose the state of your supply chain on a given day, and take operational actions to save your butt, next time we’ll talk about some reasonable goals for supply-chain visibility systems and how they can help your business.
CamiApp 