A decades-old manufacturing firm in the cement industry probably isn’t the first company that comes to mind when you think of digital transformation. But JK Lakshmi Cement has revolutionized the efficiency of its operations by using machine learning and AI.
We sat down with Avdhesh Babu, Senior Manager, Data Science & Analytics to learn more about how Dataiku has worked with their team on over 15 use cases, helping to reduce time spent on operational tasks by 60-70%.
Prior to implementing Dataiku, Avdhesh’s team was bottlenecked by the company’s lack of data- and tech-savviness. Just a few people were tasked with building reports for the whole organization, limiting the number of reports that could be created.
After starting to use Dataiku, the data team increased their production of regular reports by over 300%, transforming teams across the company. Improving the efficiency and number of reports was the first step towards transforming; the next was to grow a data culture within the organization, and then move beyond reports and automation to new and exciting AI use cases.
Building a Data-First Culture
Before Avdhesh’s team began working with Dataiku, he was on a mission to find a tool that could help them build machine learning use cases and improve their operational efficiency.
The team was experiencing two common pain points: firstly, they had scarce and underutilized data experts. Second, their data processes (at the time) lacked operationalization and the ability to make a strong business impact. The team was looking for a platform that could both boost the efficiency of its coders and allow for cross-team collaboration with line-of-business users.
Some platforms Avdhesh came across were heavily targeted toward the data scientists, but lacked flexibility for low-code and no-code users. After evaluating several other solutions on the market, he realized Dataiku could not only meet that need, but also replace many of their existing tools and integrate with their other systems. Once Dataiku was in place, he set to work on internalizing their whole data funnel, from raw data to innovative dashboards and reports.
After speeding up the time-to-delivery for reports, Avdhesh started seeing another, even more scalable benefit. Colleagues from other teams started asking for licenses so they could get their hands on Dataiku and work on their own reports. These people weren’t data scientists, but after some initial training, they were able to easily analyze their data and use it to create their own machine learning models. This allowed the core team to expand around the organization with citizen data scientists from different parts of the business, all using Dataiku to improve operational efficiency.
Creative Solutions in a Traditional Field
The flexibility and collaborative functionality of Dataiku has allowed JK Lakshmi to explore solutions to common industry pain points in novel and creative ways. In two particularly important problem spaces — customer churn and delivery prediction — Avdhesh and his team have quickly and decisively delivered improvements that have benefited the company.
One of the first use cases the team launched after starting with Dataiku was a report on which customers would stop buying from JK Lakshmi in the next couple of months. Before creating this report, the sales and marketing team had no idea where they should focus their efforts to keep more customers doing business with them. By using machine learning to predict which dealers are likely to churn, they can focus their activities to retain these customers instead of needing to re-acquire them.
Customer churn is often thought of in the context of more tech-focused industries, such as SaaS businesses, or subscription-based services like Netflix; but JK Lakshmi has demonstrated that even in a traditional industry, AI can be used to predict customer behavior and improve profitability.
Another very successful use case has been predicting delivery time for customers when they place an order. Prior to implementing Dataiku, the data team had no way of determining how long it would take to deliver their orders, and customers were left in the dark.
By using data collected from their various source systems — such as order time, unloading point, district, distance, brand, invoice time, delivery time, check-in time, gate-in time, gate-out time, and quantity — JK Lakshmi are now able to give customers near-perfect predictions of when their orders will be delivered at the time of invoicing. This has greatly improved customer satisfaction and increased trust in the data science team across the organization.
Dataiku As A Central Data Platform
One of the biggest challenges Avdhesh set out to overcome when looking for a solution was the difficulty in pulling data from their source systems. This was being done manually and then converted into the analytics tools they were using, which was one of the main reasons reports were being generated very slowly.
During the onboarding process the team at JK Lakshmi collaborated with Dataiku’s support team to build customized plugins to achieve the integration needed with the organization’s existing systems. With the ability to pull data from so many sources into Dataiku, Avdhesh’s team now has a central place to come in and work instead of needing to go to multiple places to fetch the data and manually export it. With Dataiku as a fully integrated platform JK Lakshmi was able to:
- Import and use data from multiple sources with increased efficiency.
- Spread model development to non-coder business users across the company.
- Reduce the time-to-delivery for reports by over 500% without adding additional data scientists to the team.
- Provide their core data science team more time to develop innovative solutions, enabling a move away from tedious and unnecessary manual work.
Looking Forward To Continuing Innovation
Avdhesh says that his next goal is to expand the number of Dataiku users across different teams. He continues to receive inquiries from different parts of the organization that have heard about the successes with the platform and want to try the tool for themselves.
The data team will also continue to build new use cases collaboratively with line-of-business teams, the latter being pivotal to understanding where and how the data science team can make the greatest business impact. Avdhesh is proud that JK Lakshmi has been one of the first to adopt AI driven data analysis in the manufacturing industry in India. Given the high value already generated on use cases with Dataiku, he views the ability to optimize operational performance with Dataiku as a competitive edge in an industry that often lags behind technologically.