Diagnosing A Sales Decline for a Board Game Company with Regression
Background
When a beloved board game company in Paris faced a sudden and unexpected dip in sales, their team was struggling to understand what went wrong. With a global customer base and a range of products, the root cause wasn’t immediately obvious — was it shifting customer behavior, poor timing, pricing issues, or something else entirely? They needed more than guesswork or surface-level KPIs; they needed a deep, analytical diagnosis that could turn uncertainty into action. That’s where I came in.
Challenges
The project faced several key challenges. The sales data was complex and influenced by many overlapping factors, including seasonality, product launches, and marketing campaigns. Additionally, external events—like viral reviews or social media sentiment—were hard to quantify and incorporate into traditional models. Finally, the company’s marketing and sales data were siloed, requiring careful integration and cleaning to create a cohesive dataset for analysis.
Solution
I built a regression model in R to identify and quantify the key drivers behind the sales decline. After collecting and preparing historical data — including sales trends, product launches, marketing spend, and seasonality — I conducted exploratory analysis to detect anomalies and shifts. The model revealed several unexpected contributors to the slump.
Most notably, one of the company’s flagship games had suffered a sudden drop in performance, which could be traced back to a widely viewed (and brutally honest) YouTube review that gained traction during the sales dip. At the same time, digital ad performance was also faltering — increased spend was no longer driving proportional sales, suggesting a need to reexamine targeting and channel strategy.
Impact
Armed with these insights, the company reallocated ad spend away from underperforming platforms, pulled back promotion on the criticized product, and doubled down on higher-rated titles in their portfolio. They also began actively monitoring online reviews and influencer feedback as part of their marketing analytics. Within five months, sales increased by 18%, directly tied to the corrective actions informed by the regression model. Beyond the revenue rebound, this project helped the team realize how external sentiment and timing can directly impact sales — and how data science can illuminate those hidden levers.