Our client, a global media and advertising agency serves as the principal creative partner for a world-leading FTSE100 motor company.
BI:PROCSI responded to our client’s needs during the COVID-19 pandemic by creating a Data Science MVP to forecast leads for the top EU countries and nameplates with over 70% accuracy. Leveraging advanced machine learning and Looker customisation, we provided our client with actionable insights in a user-friendly interface, enabling them to make informed decisions and drive business growth.
he challenge presented was a multifaceted endeavour, entailing the development of a predictive model capable of discerning nuanced behavioural patterns among potential automotive customers. This model needed to deliver actionable insights by forecasting leads for the “next 3 months” with an accuracy surpassing 70%, thereby enabling the assessment of conversion rates to sales and their consequential impact on the bottom line. Additionally, there existed a pressing need to scale this solution beyond its initial Minimum Viable Product (MVP) status to encompass multiple EU countries and nameplates – a task laden with complexity and multifaceted considerations in the dynamic automotive industry landscape.
The onset of the COVID-19 pandemic in 2020 disrupted the automotive industry profoundly. Initially, lead generation was significantly declining, coinciding with the imposition of lockdown measures worldwide. However, by mid-2020, leads began to rebound, surpassing pre-pandemic levels for the remainder of the year. In 2021, while leads remained below 2019 levels, they did not follow the upward trajectory seen in 2020, highlighting the enduring impact of the pandemic on automotive market dynamics.
In response to the challenges posed by the Covid-19 pandemic, BI:PROCSI embarked on a project to develop a robust solution for our client in the EU, aimed at predicting leads for potential FTSE100 automotive clients customers across various European countries and nameplates. Our data scientists analysed extensive datasets, focusing on the five Passenger Vehicle Nameplates of the EU Big 5 Countries, which generated the bulk of leads from January 2016 to August 2021. While exploring potential predictive models, we encountered limitations in data availability for newer products like MACH-E in Norway, prompting us to refine our scope.
We adopted a nuanced approach to address the complexities of predicting leads amid pandemic-related market fluctuations. We split the data into pre-pandemic (January 2016 to December 2019) and post-pandemic (January 2020 to August 2021) timelines, acknowledging the distinct behavioural patterns evident in each period. Leveraging machine learning algorithms, we identified the top 10 Countries/Nameplates combinations capable of forecasting leads with over 70% accuracy for the subsequent three months. Notably, our scenario-based modelling yielded significantly higher accuracy rates compared to models trained on the complete dataset, underscoring the importance of context in predictive analytics.
The implementation of these projects yielded transformative outcomes for the client. The modern data platform facilitated efficient centralisation and processing of large data volumes, empowering the client with comprehensive insights into customer behaviour and preferences. Furthermore, the migration to Looker augmented visualisation capabilities, enabling real-time insights and data-driven decision-making.
The recommendations engine revolutionised the customer experience by providing personalised product suggestions, fostering engagement, and driving sales.
These initiatives positioned the client as a frontrunner in the e-commerce fashion industry, enabling the brand to deliver exceptional customer experiences and foster sustainable growth in a fiercely competitive market landscape.
Third Floor, 1 London Square,
Guildford, GU1 1UN
Third Floor, 1 London Square,
Guildford, GU1 1UN
