Everyone is talking about automation these days. You have probably heard of Robotic Process Automation, Digital Transformation, AI, Blockchain, IoT, and many other tech themes. But what is the ROI of leveraging this kind of technology? Is it actually valuable, or a way for vendors like Microsoft to grow your invoice every month?
Dow Chemical as a Case Study in AI
Dow Chemical recently shared information on an automation project focused on improving tariff reporting. Dow has 150,000 shipments per year that must be classified using codes from the Harmonized System (HS). To do tariff classification, an analyst has to have an understanding of the components in the materials, the properties of the materials, regulations of the materials, and many other factors. Dow utilized the documents their analysts had generated for years, to train an AI system capable of labeling shipments with the correct HS code.
If we are generous and assume that the average analyst at Dow can classify a shipment in 5 minutes, Dow will need to dedicate 12,500 man-hours a year, solely to classification. However, the errors also come with a penalty. Dow estimated that 30% of their classifications had errors. For the sake of simplification, we will assume customs will apply a penalty of 40% of the shipment value for each incorrect filing. But realistically, we can assume customs will only notice the error 1 out of 1000 times. We will fix the average dollar value of a Dow shipping container at $38,000, based on insurance information. For the sake of simplicity, we will underestimate and assume Dow only ships one container per shipment
The full-time employee hours per year is 2080 hours, according to the US Office of Personel Management. The current task of tariff classification, therefore, requires 6 analysts. Assuming this task is given to junior team members, each would cost $53,000 a year, ignoring benefits. That means $318,000 of personnel costs.
Tariff classification could cost Dow nearly $1 Million per year between manpower and fines. We’ll assume 80% of the cases are easy enough to be automated because of the 80/20 rule. We will also conservatively assume that errors were reduced by half and the cost to maintain the automation is $100,000 a year. That still yields nearly $650,000 in savings per year. Even if we assume they had a sizable implementation budget of $500,000. That is still a 650% ROI on a 5-year time horizon.
You don’t have to be Dow Chemical to Benefit
Most firms simply do not have the operational scale of Dow Chemical. However tens of thousands of dollars per year are spent on repetitive tasks that could be automated such as tariff classification. Like Dow Chemical, the information necessary to build automated systems already exists but is likely trapped in websites, documents, and siloed datasets that enterprises don’t have the tools to leverage. For the longest time, only the largest enterprises had access to cutting edge technology by spending large sums on innovation hubs and data science teams. Companies such as Constellation are making this technology more affordable for all enterprises.
To our Supply Chain community: we’re curious to hear about your experiences in automating manual processes, especially those involving documents. Please get in touch at firstname.lastname@example.org if you have a point of view to share.
Constellation is a platform to integrate, link, and enrich enterprise data; to learn more, visit our website.