Case study
Marketing analytics automation
Weekly mobile-app performance reporting that collects, processes, and delivers itself — with AI commentary
Project overview
Our client's marketing team was assembling the same mobile-app performance report by hand every week. We built an automation that collects daily app metrics for both iOS and Android, processes them with Python, adds AI-written commentary on the trends, and delivers everything to Slack — while maintaining historical records in Google Sheets and a ready-made deck in Google Slides. The team gets two hours back every week and starts each Monday with the numbers, the trends, and the slide deck already done.
The challenge
Problem: The same mobile-app performance report was assembled by hand every week — pulling metrics, updating the tracking sheet, and building the slide deck. Two hours of repetitive work per cycle, and the numbers arrived as raw data with no commentary on what they meant.
Technical implementation
Multi-platform data collection
An automated webhook collects daily active users and app installs data from Posthog, covering both iOS and Android.
Python data processing
A Python script cleans the data and converts it to JSON, which is then uploaded to a Google Sheet tracking weekly performance.
AI commentary on the numbers
The weekly aggregated overview table, combined with the marketing activity log, is sent to an LLM to generate written commentary and identify trends — so the report explains the numbers, not just states them.
Automated distribution
The system publishes the commentary to Slack and generates a structured Slides presentation, with segmented insights across dedicated slides and graphs that visualise weekly trends.
Key features
- Runs automatically every Monday, so the marketing team starts the week with fresh numbers
- The marketing activity log adds qualitative context, so commentary reflects what the team actually did
- AI-written commentary and trend analysis with every report
- A head start on the weekly marketing slide deck, generated automatically
Need a system like this?
Tell us where work gets stuck — in one call you'll hear what it takes to remove it.
Impact
Technologies used
Business value
- Two hours of manual reporting removed every week
- Everyone sees the same numbers at the same time, every Monday
- Commentary and recommendations included, not just raw data