Data and AI
Kramp is a distributor of spare parts for the agriculture, forest and grass care, and construction industries. The family-owned company was founded in 1951 in Varsseveld, a Dutch town still home to its headquarters. With over 3,000 employees, B2B annual sales of one billion euros (in 2022), 11 warehouses, and operations in 24 countries, Kramp is Europe's largest supplier of spare parts for the above industries. Over the past 70-plus years, Kramp has consistently pursued one goal: to make work as easy as possible for its customers.
Kramp Hub, the Kramp Group's digital agency, was launched in 2017 to create solutions that move the ag-tech industry in a data-driven and customer-centric direction. Solutions include Kramp's e-commerce platform that facilitates a broad product range in the webshop and mobile app for its dealers, and Maykers, Kramp's marketplace for the farmers.
Becoming a Data-Driven Organization
Kramp's ambition is to be a digital leader and a data-driven organization. The company approached Xebia Data to help with the latter, as Xebia Data is uniquely positioned to both build a data strategy and bring it to fruition.
Creating a Data Strategy
Creating a Digital Strategy
Steven Nooijen, Head of Data and Strategy at Xebia Data, was of great value to Kramp in shaping its data strategy. The process they worked through together consisted of the following four components:
I. Maturity assessment: where are we today in terms of data, people, tools, and processes?
II. North Star goal setting: three to five years from now, what does being successful with data look like? How should data contribute to our goals? And, moreover: Integrating the Data strategy into the overall group strategy.
III. Use case ideation: which data and AI use cases yield business value? Collect, refine, and prioritize use cases across all business divisions.
IV. Strategy creation: what must we do to get from where we are today to where we want to be in three years?
Based on the maturity audit & workshops, Xebia Data concluded five different tracks to define the strategy: more focus on data quality, professionalizing of enterprise data warehouse system, building a self-service data science/MLOps platform to enable data scientists to develop models, establishing a data Office (DCoE) and hiring a data director, and finally, talent development through training and clear growth paths.
Part of the strategy was a cost-benefit analysis, in which we estimated the expected ROI (from conservative to progressive).
Bringing the Data Strategy to Life
The new strategy was followed up with the implementation of an MLOps platform and bringing the first use case to life on the platform by Xebia's Data Engineering, Machine Learning Engineering, and Data Science consultants in collaboration with Kramp's in-house team.
Eva Bosma, Analytics Translator at Xebia Data, was in charge of defining and refining use cases with Kramp. Her workshops led to 14 use cases, ranging from demand forecasting — so that operations can optimize capacity planning — to creating a dashboard with customer and product profitability insights.
Today, Kramp has a brand-new MLOps platform on GCP, using Vertex AI and Terraform, and a dedicated team that will work with it — a great foundation to monetize many use cases.
"The first use case that we selected is the Demand Forecasting use case. The goal is to improve the current forecasting model to increase our product availability and decrease stock costs. We have delivered the first version of the model (only based on historical data) and we already see a huge improvement compared to our old model. The next step to improve and enrich the forecasting model is to include customer behaviour data from our e-commerce platform, seasonal influences, and other interesting external data sources. This is all possible thanks to the platform and the knowledge we now have thanks to Xebia Data." — Chiel Schutter, Director of Kramp Hub.
Based on the implemented model, the current forecast has been improved by 4%. Compared to the previous model, the recently implemented model will result in more efficient stock management and availability. Kramp expects that with this new model, cost savings of around 8 million EURO will be realized.