Smartphone Pricing & Demand Analytics

The Problem

Phone makers can't afford to guess at prices. Price a device too high, and you lose customers to competitors. Price it too low, and you leave profit on the table. Companies also tend to treat all customers the same, when in reality different people care about price, brand, and features very differently. For example, price sensitivity can vary more within a company's own customers than between competitors.

What I Did

Using purchase data from thousands of smartphone customers, I grouped customers into segments based on how they actually use their phones (social media, gaming, video, etc.). I then built statistical models that predict which phone a customer will choose based on price, brand, and their personal characteristics. I tested a few versions of this model and used cross-validation to pick the one that actually predicted customer choices best.

The Results

The models showed that price sensitivity varies meaningfully across customer segments, as some groups are much more price-driven than others. Using the best-performing model, I simulated different pricing scenarios for two Huawei phones and found a combination that would increase profit by about 13% over current pricing. I also forecasted category-wide and brand-specific sales three years out using a diffusion model.

What's Next

With more time, I'd incorporate competitor reactions (would rivals cut prices in response?) and test the pricing recommendations against a holdout period of real sales data rather than just simulated demand, to see how well the model's profit predictions actually hold up.