# SpotLift
> Python-native econometric library for minute-by-minute Linear TV attribution across Google Analytics 4 (GA4) and Amazon Marketing Cloud (AMC).

## Core Capabilities
- **Open-Source Core:**
  - 21-day diurnal rolling median baseline de-trending with 15-minute pre-spot level calibration.
  - Minute-level single-spot incremental lift integration for GA4 sessions and conversions.
  - Automated overlapping spot collision detection across broadcast channels.
  - Unbiased signed lift integration (`clip_negative=False`) preventing noise rectification bias.
- **SpotLift Pro (`spotlift.install_pro(api_key)`):**
  - Expectation-Maximization (EM) multi-spot deconvolution using Shifted-Gamma impulse response kernels.
  - Dual-stage TV-to-Amazon Distributed Lag attribution (linking immediate search spikes to 1–36h delayed unit orders).
  - Amazon Branded PPC Cannibalization Auditor (separating genuine TV organic halo from Sponsored Brands click interception) + AMC Clean Room SQL generator.
  - Automated Post-Buy Efficiency Matrix (Station, Daypart, Program Genre, Pod Position, Creative Wear-out) exported to interactive HTML and Excel.

## Quickstart
```python
import spotlift as sl

airings = sl.load_airings("tv_schaltplan.csv")
ga4 = sl.load_telemetry("ga4_minute_sessions.parquet")

# Free baseline & single-spot attribution
report = sl.analyze_spots(airings, ga4, window_minutes=8)
```

## Links
- Repository: https://github.com/flovv/spotlift
- Documentation: https://flovv.github.io/spotlift/
- Docs Source: https://github.com/flovv/spotlift/tree/main/docs
- Agent Skill: https://github.com/flovv/spotlift/tree/main/skills/tv-to-web-measurement/SKILL.md
