Organizations are exploring how to integrate digital transformation, and find ways to exploit emerging technologies, such as AI/ML/DL, IoT, Data analytics, BlockChain, Cloud ERP, and more. Digitalization cannot be obtained over the counter, link it to ERP, and begin running analytics applications. The shortage of good people, particularly data scientists, is the real bottleneck of deploying emerging technologies. Moreover, experimentation through pilot projects is key to blaze the path ahead. The main problem is how to determine an ROI on a sizeable investment in technologies that are still in the cutting-edge stage. Not having measures for a clear ROI will hinder initiatives in investing in these emerging technologies. However, measures cannot be limited to only improved cost and efficiency. Measures can involve dimensions that customers will value. For example,increased speed of decision-making, increased responsiveness, and higher performance analytics that are able to forecast accurately current and future needs are all of great value for customers.

     The second major problem that slows or hinders emerging technologies adoption and deployment is good data. Granted, there is data everywhere and coming torrentially from various sources. However, AI/ML/DL, IoT, and BlockChain deployment require good data. Good Data is paramount to any decision-making. To be trusted, measures of data include accuracy, completeness, timeliness, consistency, and uniqueness. Today, a large majority of organizations (up to 80%) believe their data cannot be trusted to deploy AI/ML/DL. Managers across industries use analytics for visualizing trends, status of their finance, sales, productivity, and more in the form of graphs, charts, or other representations. They assume the data used is a good fit to business reality. Hence, they take decisions based on these assumptions.

     Management of emerging supply chain technologies is completely different from management of “old” technologies of the past such as EDI, ERP or even Cloud-ERP. They are deployed as running complete systems. True, they were in many initiatives implemented as pilot projects to experiment and learn. However, they come as a complete system.

The emerging technologies is inherently dependent on good data. They are deployed to take advantage of the torrent of data and Big Data. Machine Learning algorithms are useless if trained on corrupt data. Experimentation is key to create a kind of analytics culture. ML (supervised and unsupervised) algorithms can be used to clean the data and allow to exploit and experiment rapidly. Therefore, ML algorithms should be deployed and used to automatically clean data and make data quality available for “trainable” ML applications. For example, to develop a reliable classification and clustering models, there are various supervised machine learning algorithms to classify and unsupervised algorithms to cluster for data cleansing. Data scientists can use clustering algorithms to detect outliers (an outlier is a data point that differs significantly from other observations) and “dirty” data. Classification algorithms (which are supervised ML) can be applied to assign the correct class label or detect duplicate data. Likewise, deep learning Convolutional Neural Network (CNN) can be deployed to generate potential domains. Cleaning is not highlighted too often. Yet, it is critical for quality analytics and prediction. Powerful algorithms without quality data training will yield wrong prediction and prescription. The bottleneck for organizations is data scientists’ talent. Data scientists are hard to find and to train. New emerging technology-based supply chain requires the training of scientists with a new culture that think beyond “cost and efficiency metrics.” 

Summing-up:  Measures and ROI can slow or hinder supply chain emerging technologies. The scarcity of finding good people to experiment and exploit these technologies is a major barrier for technology adoption. Considering the challenges ahead, organizations should hire and start training data scientists capable of creating a new culture, that think beyond “cost and efficiency metrics.” Moreover, organizations would consider applications that will speed-up decision-making, increase asset velocity, and create new performance analytics in real-time that enhance visibility, trust and mutual understanding between trading partners, and other new capabilities that are of important value to customers.