← Back to case studies
Data & Market Research
Data & Market Research: Revenue Benchmarking
Value
15x speed gain
Timeline
4-6 weeks (each)
The Problem
Current practices only considered basic data for revenue estimation. These sources had either bias towards platforms or industries or lacked data on platforms.
The Solution
Multi-model approach: a) Created a rule-based blend model to join multiple sources to improve revenue estimation downsteam and its robustness, b) Pilot an app-based revenue estimation model, c) Build a scalable, robust, parallelized web-scraper for app-related data, d) Validate and improve current in-house modeling techniques by benchmarking against state-of-the art models.
Tech Stack
PythonPostgreSQLPower BI
The Impact
90%
90% of platforms covered by new blending method
15
15x faster web scraper
40%
40% of forecasts with improvement