Sage

Image credit: Canva
The company
Sage, a global software company headquartered in Newcastle upon-Tyne, provides accounting, customer relationship management (CRM), and enterprise resource planning (ERP) solutions to businesses worldwide.
Sage aims to knock down barriers so everyone can thrive, starting with the millions of small and medium-sized businesses they serve. The National Innovation Centre for Data (NICD) collaborated with Sage to enhance its internal data science capabilities and provide innovative data science expertise.
The problem
Like many companies with a subscription-based business model, Sage identified a challenge with customer churn (the number of people who stopped using a product or service during a set period). The company wanted to minimise churn to maximise the value of their customer base.
An innovative Marketing Data Science Team was established within Sage in August 2021, consisting of mostly junior data scientists and analysts, who needed guidance to tackle this challenge effectively.

The goal
The primary goal was to develop a robust churn prediction model and design appropriate intervention strategies. This involved upskilling the newly created Marketing Data Science Team to ensure they had the necessary skills to manage and enhance the model independently.
The solution
NICD supported the Sage team through a comprehensive training program covering key data science concepts and tools.
Sam Urwin is a Data Scientist at Sage, “NICD helped with the whole process end-to-end, from building and deploying a machine learning model to showing us how to do things properly.”
Sam, alongside the Sage team collaborated with NICD Data Scientists Dr Mac Misiura and Dr Matt Edwards who provided training on various data science tools and techniques, including:
- Data cleaning and pre-processing: Emphasising the importance of data quality, crucial for building accurate models.
- Classification models: Focused on tree-based models, effective for classification tasks.
- Dealing with imbalanced data: Techniques to handle imbalanced datasets, ensuring the models could effectively predict churn.
- Model explainability: Support on how to explain complex machine learning to stakeholders, allowing the team to communicate effectively. This included an introduction to Shapley Values and SHAP for visualising and explaining model predictions.
- End-to-end machine learning pipeline: From data collection and cleaning to model deployment and monitoring, ensuring a comprehensive understanding of the entire process.
Image credit: Canva
Implementation
The NICD team provided hands-on training through a mixture of formal presentations, pair programming sessions, and ongoing support. “They walked us through the whole process,” said Sam. “From looking at different data sources, data cleaning, developing the model, to implementing and monitoring it.”
With the project lasting two years, there was a steady transfer of skills to the new team members– which allowed the Sage team to apply their learning throughout to real-world problems.
"It has been a real pleasure working with the data scientists at Sage. As the project progressed from Churn Prediction to the entire end-to-end Machine Learning pipeline, I was able to share virtually every tip and trick I have learnt from the last five years working with organisations on ML (Machine Learning) projects.
As the project progressed it was great to see these learnings applied in practice and the results that followed."
Dr Matt Edwards, Senior Data Scientist, NICD
The result
The collaboration with NICD led to the development of a prediction model that allowed Sage to identify customers at high risk of churn earlier. This allowed help and guidance to be offered with a view to delivering a better customer experience and ultimately reducing churn.
The project also resulted in significant knowledge transfer, empowering the Marketing Data Science Team to handle other projects independently. Estimations from initial tests of another churn prediction model indicated significant cost saving through the reduction of churn for their Sage Business Cloud Payroll product. “The knowledge gained from that initial project has helped us to develop subsequent projects requiring end-to-end machine learning processes,” noted Sam. “It would be hard to quantify anything further because the impact has been so wide ranging. This project has impacted many areas of our work.”
Sam shared another specific example of how the skills gained were applied later in the project lifecycle, “Recently, we got access to a major data vendor’s platform, which was hosted on Databricks. Thanks to the training from NICD, we were able to get up to speed quickly and develop macro churn and migration models across multiple products and regions.”

Image credit: Sage
Business impact
The immediate impact of the project was substantial, with a significant reduction in churn and a corresponding cost saving. Beyond this direct fiscal impact, the knowledge and skills gained through the project have been applied across other products and regions within Sage. “Collaborating with NICD and sharing specific examples of the code made learning quicker and more valuable,” mentioned Sam.
The collaboration fostered a strategic partnership between Sage and NICD, contributing to the onboarding and upskilling of newly hired data scientists and analysts.
“I am thrilled to have guided Sage on their data science journey, witnessing their team grow and skills soar. Now, they are equipped with the expertise and confidence to tackle future challenges head-on.”
Dr Mac Misiura, Data Scientist, NICD
Empowering future success
The partnership between Sage and NICD has created a lasting impact. The dedicated time and collaborative approach fostered a deep transfer of knowledge, allowing the Marketing Data Science Team to excel in their roles. As Carl Mills, a Data Scientist at Sage, highlighted, “The dedicated time, collaboration on projects, learning as you go, trying out different things approach was far more effective than online training courses.” This has led to substantial improvements in the ability of Sage to handle complex data science projects independently.
For Carl, the expert guidance proved invaluable: “Getting NICD’s opinion on a gold standard to then implement was extremely beneficial, and we will continue to use the knowledge gained on future projects.” The training and ongoing support provided by NICD have been invaluable, ensuring the Sage team stays updated with the latest trends in the fast-evolving field of data science.

Image credit: Sage
To find out more about Sage, visit their website.
You can read more of our case studies and sign up to our newsletter to keep up to date with our latest news, events and developments.

Our Discovery workshop
Our Discovery workshops enable you to explore the potential of your data and understand the benefit you could gain before committing to a full-scale project.