Strategy Consulting
Roland Berger
Project Consulting Intern
May – Aug 2025Wuhan, Hubei, China
Embedded in Roland Berger's automotive user-growth practice, I led quantitative analysis for a flagship engagement with Dongfeng Honda — helping the client translate three years of KOC ecosystem data into a deployable growth strategy.
R² = 0.86
Model accuracy
500+
User profiles clustered
27%
UGC output increase
97 pages
Operation system manual
Data Processing
- Integrated 3 years of sales and user operations data for Dongfeng Honda across 10 dimensions including KOC engagement, content creation, and referral rates.
- Used Python (pandas, NumPy) to clean and validate data, detect anomalies, and identify key growth factors to support the brand's private-domain growth strategy.
Model Development
- Built multiple linear regression and random-forest models to forecast new KOC growth and content output trends.
- Achieved R² = 0.86, directly informing quarterly operation targets and resource allocation decisions.
- Produced 10+ slide decks summarising model results — all adopted by the client.
Strategy Optimisation
- Co-authored multiple rounds of executive-level presentation materials and designed KOC growth paths and UGC incentive tiering systems.
- Applied K-Means clustering to 500+ user profiles and content types, identifying three core segments: "Aesthetic Seeding," "Daily Sharing," and "Usage Insights."
- Updated the KOC grading system using the Diamond Model and proposed differentiated strategies by vehicle type and region — raising UGC output by 27% in the following month.
System Building
- Led creation of the 97-page Dongfeng Honda KOC Digital Operation System Manual (PowerPoint), converting project insights into a standardised operational framework.
- Produced a 15-page Word standard operation guide for internal training and process standardisation.
Tools & Stack
PythonpandasNumPyK-MeansRandom ForestMultiple RegressionPowerPointExcel