Skip to content
Advertisements

Posts tagged ‘McKinsey Global Institute’’

How Machine Learning Improves Manufacturing Inspections, Product Quality & Supply Chain Visibility

Bottom Line: Manufacturers’ most valuable data is generated on shop floors daily, bringing with it the challenge of analyzing it to find prescriptive insights fast – and an ideal problem for machine learning to solve.

Manufacturing is the most data-prolific industry there is, generating on average 1.9 petabytes of data every year according to the McKinsey Global Insititute. Supply chains, sourcing, factory operations, and the phases of compliance and quality management generate the majority of data.

The most valuable data of all comes from product inspections that can immediately find exceptionally strong or weak suppliers, quality management and compliance practices in a factory. Manufacturing’s massive problem is in getting quality inspection results out fast enough across brands & retailers, other factories, suppliers and vendors to make a difference in future product quality.

How A Machine Learning Startup Is Revolutionizing Product Inspections

Imagine you’re a major brand or retailer and you’re relying on a network of factories across Bangladesh, China, India, and Southeast Asia to produce your new non-food consumer goods product lines including apparel. Factories, inspection agencies, suppliers and vendors that brands and retailers like you rely on vary widely on ethics, responsible sourcing, product quality, and transparency. With your entire consumer goods product lines (and future sales) at risk based on which suppliers, factories and product inspection agencies you choose, you and your companies’ future are riding on the decisions you make.

These career- and company-betting challenges and the frustration of gaining greater visibility into what’s going on in supply chains to factory floors led Carlos Moncayo Castillo and his brothers Fernando Moncayo Castillo and Luis Moncayo Castillo to launch Inspectorio. They were invited to the Target + Techstars Retail Accelerator in the summer of 2017, a competition they participated in with their cloud-based inspection platform that includes AI and machine learning and pervasive support for mobile technologies. Target relies on them today to bring greater transparency to their supply chains. “I’ve spent years working in non-food consumer goods product manufacturing seeing the many disconnects between inspections and suppliers, the lack of collaboration and how gaps in information create too many opportunities for corruption – I had to do something to solve these problems,” Carlos said. The many problems that a lack of inspection and supply chain visibility creates became the pain Inspectorio focused on solving immediately for brands and retailers. The following is a graphic of their platform:

Presented below are a few of the many ways the combining of a scalable inspection cloud platform combined with AI, machine learning and mobile technologies are improving inspections, product quality, and supply chain visibility:

  • Enabling the creation of customized inspector workflows that learn over time and are tailored to specific products including furniture, toys, homeware and garments, the factories they’re produced in, quality of the materials used. Inspectorio’s internal research has found 74% of all inspections today are done manually using a pen and paper, with results reported in Microsoft Word, Excel or PDFs, making collaboration slow and challenging. Improving the accuracy, speed and scale of inspection workflows including real-time updates across production networks drive major gains in quality and supply chain performance.
  • Applying constraint-based algorithms and logic to understand why there are large differences in inspection results between factories is enabling brands & retailers to manage quality faster and more completely. Uploading inspections in real-time from mobile devices to an inspection platform that contains AI and machine learning applications that quickly parse the data for prescriptive insights is the future of manufacturing quality. Variations in all dimensions of quality including factory competency, supplier and production assembly quality are taken into account. In a matter of hours, inspection-based data delivers the insights needed to avert major quality problems to every member of a production network.
  • Reducing risk, the potential for fraud, while improving the product and process quality based on insights gained from machine learning is forcing inspection’s inflection point. When inspections are automated using mobile technologies and results are uploaded in real-time to a secure cloud-based platform, machine learning algorithms can deliver insights that immediately reduce risks and the potential for fraud. One of the most powerful catalysts driving inspections’ inflection point is the combination of automated workflows that deliver high-quality data that machine learning produces prescriptive insights from. And those insights are shared on performance dashboards across every brand, retailer, supplier, vendor and factory involved in shared production strategies today.
  • Matching the most experienced inspector for a given factory and product inspection drastically increases accuracy and quality. When machine learning is applied to the inspector selection and assignment process, the quality, and thoroughness of inspections increase. For the first time, brands, retailers, and factories have a clear, quantified view of Inspector Productivity Analysis across the entire team of inspectors available in a given region or country. Inspections are uploaded in real-time to the Inspectorio platform where advanced analytics and additional machine learning algorithms are applied to the data, providing greater prescriptive insights that would have ever been possible using legacy manual methods. Machine learning is also making recommendations to inspectors on which defects to look for first based on the data patterns obtained from previous inspections.
  • Knowing why specific factories and products generated more Corrective Action/Preventative Action (CAPA) than others and how fast they have been closed in the past and why is now possible. Machine learning is making it possible for entire production networks to know why specific factory and product combinations generate the most CAPAs. Using constraint-based logic, machine learning can also provide prescriptive insights into what needs to be improved to reduce CAPAs, including their root cause.
Advertisements

McKinsey’s 2016 Analytics Study Defines The Future Of Machine Learning

  • U.S. retailer supply chain operations who have adopted data and analytics have seen up to a 19% increase in operating margin over the last five years.
  • Design-to-value, supply chain management and after-sales support are three areas where analytics are making a financial contribution in manufacturing.
  • 40% of all the potential value associated with the Internet of Things requires interoperability between IoT systems.

These and many other insights are from the McKinsey Global Institute’s study The Age of Analytics: Competing In A Data-Driven World published in collaboration with McKinsey Analytics this month. You can get a copy of the Executive Summary here (28 pp., free, no opt-in, PDF) and the full report (136 pp., free, no opt-in, PDF) here. Five years ago the McKinsey Global Institute (MGI) released Big Data: The Next Frontier For Innovation, Competition, and Productivity (156 pp., free no opt-in, PDF), and in the years since McKinsey sees data science adoption and value accelerate, specifically in the areas of machine learning and deep learning. The study underscores how critical integration is for gaining greater value from data and analytics.

Key takeaways from the study include the following:McKinsey Analytics

  • Location-based services and U.S. retail are showing the greatest progress capturing value from data and analytics. Location-based services are capturing up to 60% of data and analytics value today predicted by McKinsey in their 2011 report. McKinsey predicts there are growing opportunities for businesses to use geospatial data to track assets, teams, and customers across dispersed locations to generate new insights and improve efficiency. U.S. Retail is capturing up to 40%, and Manufacturing, 30%.  The following graphic compares the potential impact as predicted in McKinsey’s 2011 study with the value captured by segment today, including a definition of major barriers to adoption.

uneven-progress

  • Machine learning’s greatest potential across industries includes improving forecasting and predictive analytics. McKinsey analyzed the 120 use cases their research found as most significant in machine learning and then weighted them based on respondents’ mention of each. The result is a heat map of machine learning’s greatest potential impact across industries and use case types.  Please see the report for detailed scorecards of each industry’s use case ranked by impact and data richness.

machine-learning-impact

  • Machine learning’s potential to deliver real-time optimization across industries is just starting to evolve and will quickly accelerate in the next three years. McKinsey analyzed the data richness associated with each of the 300 machine learning use cases, defining this attribute as a combination of data volume and variety. Please see page 105 of the study for a thorough explanation of McKinsey’s definition of data volume and variety used in the context of this study The result of evaluating machine learning’s data richness by industry is shown in the following heat map:

rich-data-is-an-enabler

  • Enabling autonomous vehicles and personalizing advertising are two of the highest opportunity use cases for machine learning today. Additional use cases with high potential include optimizing pricing, routing, and scheduling based on real-time data in travel and logistics; predicting personalized health outcomes, and optimizing merchandising strategy in retail. McKinsey identified 120 potential use cases of machine learning in 12 industries and surveyed more than 600 industry experts on their potential impact. They found an extraordinary breadth of potential applications for machine learning.  Each of the use cases was identified as being one of the top three in an industry by at least one expert in that industry. McKinsey plotted the top 120 use cases below, with the y-axis shows the volume of available data (encompassing its breadth and frequency), while the x-axis shows the potential impact, based on surveys of more than 600 industry experts. The size of the bubble reflects the diversity of the available data sources.

machine-learning

  • Designing an appropriate organizational structure to support data and analytics activities (45%), Ensuring senior management involvement (42%), and designing effective data architecture and technology infrastructure (36%) are the three most significant challenges to attaining data and analytics objectives. McKinsey found that the barriers break into the three categories: strategy, leadership, and talent; organizational structure and processes; and technology infrastructure. Approximately half of executives across geographies and industries reported greater difficulty recruiting analytical talent than any other kind of talent. 40% say retention is also an issue.

barriers-to-analytics-and-machine-learning-adoption

  • U.S. retailer supply chain operations who have adopted data and analytics have seen up to a 19% increase in operating margin over the last five years. Using data and analytics to improve merchandising including pricing, assortment, and placement optimization is leading to an additional 16% in operating margin improvement. The following table illustrates data and analytics’ contribution to U.S. retail operations by area.

us-retail-data-sheet

  • Design-to-value, supply chain management and after-sales support are three areas where analytics are making a financial contribution in manufacturing. McKinsey estimates that analytics have increased manufacturer’s gross margins by as much as 40% when used in design-to-value workflows and projects. Up to 15% of after-sales costs have been reduced through the use of analytics that includes product sensor data analysis for after-sales service. There are several interesting companies to watch in this area, with two of the most innovative being Sight Machine and enosiX, with the latter enabling real-time integration between SAP and Salesforce systems. The following graphic illustrates the estimated impact of analytics on manufacturing financial performance by area.

manufacturing

%d bloggers like this: