Kirana Ananda

Data Science Marketing Campaign Market Segmentation

Right Message, Right Audience: Optimizing Airbnb’s Campaign with Segmentation

Using machine learning to transform an "underperforming favorite" market into a targeted growth opportunity.

✦

Role
Data Analyst

Tools
Python, Pandas, Scikit-learn

Method
Hierarchical Clustering

The Challenge

Chicago was identified as an "Underperforming Favorite." It had high distinctiveness but low demand conversion. Generic marketing wasn't moving the needle. The goal was to move beyond price-based segmentation and understand who was actually visiting based on behavior and amenities.

The Approach

Standard clustering often uses Euclidean distance (price/location). However, amenity data is binary (Has Pool? Yes/No). To solve this, I engineered a different approach:

Data Engineering: Processed 500+ raw amenities into 27 core value categories (e.g., "Remote Work Ready," "Family Friendly").
The Algorithm: Applied Hierarchical Clustering (Complete Linkage).
Why Jaccard? I chose Jaccard Distance over Euclidean because it statistically measures dissimilarity better for binary data sets.

# Calculate Distance Matrix
from scipy.spatial.distance import pdist

# Using Jaccard for binary amenity data
dist_matrix = pdist(
    amenity_df, 
    metric='jaccard'
)

# Hierarchical Clustering
model = AgglomerativeClustering(
    n_clusters=4, 
    linkage='complete'
)

The Discovery: 4 Key Personas

The model revealed four distinct traveler types that generic marketing had missed.

Backyard Bliss

39% of Market

Travelers seeking private outdoor space. They value patios, grills, and gardens over luxury interiors.

Active Leisure

High Value / Low Vol

Guests prioritizing fitness. They look for pools, gyms, and proximity to parks. Highest willingness to pay.

The Essentialist

Budget Focused

Travelers who just need a bed and Wi-Fi. Minimal amenity requirements. Price sensitive.

Group/Family

High Occupancy

Large groups needing "togetherness" features: large kitchens, multiple bathrooms, and parking.

The Solution: Seasonal Content Calendar

I translated these clusters into a targeted marketing schedule to maximize seasonal demand.

SPRING

Target: Backyard Bliss

Highlight "First days of sun." Promote listings with patios and grills before the summer rush begins.

SUMMER

Target: Active Leisure

Pivot to "Cool & Active." Feature pools, gyms, and AC to capture high-value bookings during peak heat.

FALL/WINTER

Target: Family & Groups

Focus on holidays. Promote large kitchens and "togetherness" spaces for Thanksgiving/Christmas travel.