In this blog I am going to call out the specific areas where machine learning has shown transformational results in a business context and highlight a few use cases that can benefit from leveraging machine learning.
Before we go further I would like to mention why I see machine learning becoming an important factor in business transformation leveraging analytics.
- Traditional business transformation leveraging analytics relies on definition of static business rules that need to keep up with constant changes in business.
- These are usually deeply embedded in the core analytic systems and applications and are difficult to change and thus become less effective over time.
- The above two limitations are key reasons why business productivity typically stalls after the initial deployment of the analytical solutions.
With machine learning we can have a dynamic element to refining the business rules based on patterns derived from analyzing massive amounts of data with business real-time capability which avoids the above mentioned problems.
Below are few examples of use cases where enterprises have seen significant improvements in business indicators leveraging machine learning for driving business transformation without incurring regular maintenance of business rules for accuracy of reported business intelligence information.
Use case #1
involving difficulty in separating data points of seemingly similar but semantically different data patterns (traditionally solved by human managed business rules that require rigorous ongoing maintenance and costs).
Use case #2
- Airbus Defense and Space, tested the use of Google Cloud Machine Learning to automate the process of detecting and correcting satellite images that contain imperfections such as the presence of cloud v/s snow formations. This business intelligence service that was derived from this was extensively by enterprises managing farming operations, construction operations across Europe. Using machine learning eliminated the deficiencies of the previous time consuming, error-prone solution that was unable to scale with the needs of the enterprises. Results of the improved solution were improved farming profitability due to increase crop yields, lower construction costs with the ability to predict environmental impact on construction projects.1
involving multiple drivers of demand information with or without complex offerings and multiple data management systems/applications/repositories with or without disjointed front end and back end processes in Value Chain.
- Danone achieved significant business transformation by connecting disconnected front end complex demand and forecasting processes with their operational back end planning processes while adopting predictive commerce and forecasting methods leveraging machine learning. The key to Danon’s success was improving forecast accuracy by leveraging machine learning and using upstream and downstream data from both internal/external data sources. Results of the improved solution were a direct increase in forecast accuracy to enviable 92% with decrease of forecast error by 20%. The indirect benefits of this were reduction in lost sales by 30% and reduced obsolescence of products by 30%.2
Use case involving inability to figure out drivers or having to deal with unstructured data to drive business transformation in a Retail Aspect.
- Ocado Technologies an online only retail grocery supermarket with 500,000 customers had to figure out how to service their customers better and yet improve profitability while dealing with mostly textual content. They applied machine learning to determine patterns of customer behavior that not only helped them improve the customer interaction in term of faster response to customers but also helped them make their distribution center efficient by automating the way they managed the storage and retrieval of their inventory for the most efficient distribution.3
With the above background here are a few use cases within specific industries that I would like to highlight as potential use cases where machine learning based business transformation could lead to significant business benefits.
High Technology/Discrete Manufacturing:
Services including Finance:
- Propensity to buy including improved service attach rates
- Demand forecasting
- Supply Chain optimization especially for those enterprises with complex offerings and rapid changes
- Predictive maintenance or condition monitoring
- Warranty reserve estimation
- Cross-selling and up-selling
- Customer Segmentation
- Sales and marketing campaign management
- Credit worthiness evaluation
- Risk analytics and regulation
Key factors to keep in mind for achieving success with Machine Learning driven Business Transformation
- Recommendation engines
- Upsell and cross-channel marketing
- Market segmentation and targeting
- Predictive inventory planning
- Customer ROI and lifetime value
- Managing Omni channel value chains
- Right Use Case identification
- Identification of data and quality of data; data needs to have right representation to ensure that is representative of data that is required for achieving desired business results
- Optimize the algorithm for improving prediction accuracy (use step by step approach of training of models, validation of models and selection of models to set the initial set of business rules).
- Use a crawl, walk and run approach
In my next blog I will describe how to successfully adopt the crawl, walk and run approach for leveraging machine learning for business transformation.
1 From reference information publicly shared by Google on the work done with Airbus
2 From 2015 Material Handling and Logistics Conference
3 From Case study illustrating usage of machine learning at Ocado